Confocal microscopy enables us to track myocytes in the embryonic zebrafish heart. The Zeiss LSM 5 Live high speed confocal microscope has been used to take optical sections (at 3 μm intervals and 151 frames per second) through a fluorescently labeled zebrafish heart at two developmental stages (26 and 34 hours post fertilization (hpf)). This data provides unique information allowing us to conjecture on the morphology and biomechanics of the developing vertebrate heart. Nevertheless, the myocytes, whose positions could be determined in a reliable manner, were located sparsely and mostly in one side of the heart tube. This difficulty was overcome using computational methods, that give longitudinal, radial and circumferential displacements of the myocytes as well as their contractile behavior. Applied strain analysis has shown that in the early embryonic heart tube, only the caudal region (near the in-flow) and another point in the middle of the tube can be active; the rest appears to be mostly passive. This statement is based on the delay between major strain and displacement which a material point experiences. Wave-like propagation of all three components of the displacement, especially in the circumferential direction, as well as the almost-periodic changes of the maximum strain support the hypothesis of helical muscle structure embedded in the tube. Changes of geometry in the embryonic heart after several hours are used to verify speculations about the structure based on the earlier images and aforementioned methods.

Obesity is a global epidemic and a comorbidity for many diseases. We are using MRI to characterize obesity in rodents, especially with regard to visceral fat. Rats were scanned on a 1.5T clinical scanner, and a T1W, water-spoiled image (fat only) was divided by a matched T1W image (fat + water) to yield a ratio image related to the lipid content in each voxel. The ratio eliminated coil sensitivity inhomogeneity and gave flat values across a fat pad, except for outlier voxels (> 1.0) due to motion. Following sacrifice, fat pad volumes were dissected and measured by displacement in canola oil. In our study of 6 lean (SHR), 6 dietary obese (SHR-DO), and 9 genetically obese rats (SHROB), significant differences in visceral fat volume was observed with an average of 29±16 ml increase due to diet and 84±44 ml increase due to genetics relative to lean control with a volume of 11±4 ml. Subcutaneous fat increased 14±8 ml due to diet and 198±105 ml due to genetics relative to the lean control with 7±3 ml. Visceral fat strongly correlated between MRI and dissection (R² = 0.94), but MRI detected over five times the subcutaneous fat found with error-prone dissection. Using a semi-automated images segmentation method on the ratio images, intra-subject variation was very low. Fat pad composition as estimated from ratio images consistently differentiated the strains with SHROB having a greater lipid concentration in adipose tissues. Future work will include in vivo studies of diet versus genetics, identification of new phenotypes, and corrective measures for obesity; technical efforts will focus on correction for motion and automation in quantification.

We believe that small-animal positron emission tomography (μPET) can play an important role in phenotyping and drug screening. For such applications, imaging throughput becomes an important issue because one needs to image a considerable number of subjects in a study. Toward enabling high-throughput μPET imaging, we are developing a prototype that consists of two large-area, high-performance flat detectors. These detectors are placed opposed to each other with a small spacing for providing large detection solid angle and detection sensitive volume. The resulting scanner geometry produces data having missing views and projection truncations, therefore posing a particular challenge in reconstruction. In this paper, we developed a new iterative reconstruction method that addresses this challenge. By using 2D simulated data, we find that this new method can accurately reconstruct an extended detection volume of the prototype. Because our prototype shares the same configuration with positron emission mammography (PEM), the new reconstruction method is also applicable for PEM reconstruction.

Technological advances in micro-CT scanners have introduced dynamic, flat-panel scanners, which allow the acquisition of volume images in a few seconds. However, motion artefacts associated with normal respiratory motion arise when imaging the thorax or abdomen. To reduce these artefacts and the accompanying loss of spatial resolution, and to enable the study of rodent respiratory function, we developed a retrospective respiratory gating technique for volume micro-CT imaging of free-breathing rodents. Anaesthetized male C57BL6 mice were placed in the prone position on a custom-made bed containing an embedded pressure chamber that was connected to a pressure transducer. Inhalation motion caused an increase in the chamber pressure, which was monitored as a surrogate for the respiratory waveform, and measured throughout the scan. Projection images of the mouse thorax were acquired using a GE Locus Ultra micro-CT scanner, at 80 kVp, 50 mA (entrance exposure of approximately 2.7 cGy per rotation), over ten rotations in less than 1 minute. Respiratory gating was performed retrospectively by selecting projections that were obtained during the same portion of the respiratory cycle prior to reconstruction; CT images reconstructed from three to ten rotations were evaluated. The nominal voxel spacing was 0.15 mm isotropic. Images were assessed for image noise, artefacts and measurement accuracy of physiologically relevant structures. These measurements showed no significant differences for images reconstructed from projection images from five to ten rotations. The optimum number of rotations for imaging mouse lungs was found to be six, corresponding to a 30 second (16.2 cGy) scan.

Micro-CT, a technique for imaging small objects at high resolution using micro focused x-rays, is becoming widely available for small animal imaging. With the growing number of mouse models of pulmonary pathology, there is great interest in following disease progression and evaluating the alteration in longitudinal studies. Along with the high resolution associated with micro CT comes increased scanning times, and hence minimization of motion artifacts is required. We propose a new technique for imaging mouse lungs in vivo by inducing an intermittent iso-pressure breath hold (IIBH) with a fixed level of positive airway pressure during image acquisition, to decrease motion artifacts and increase image resolution and quality. Mechanical ventilation of the respiratory system for such a setup consists of three phases, 1) tidal breathing (hyperventilated), 2) a breath hold during a fixed level of applied positive airway pressure, 3) periodic deep sighs. Image acquisition is triggered over the stable segment of the IIBH period. Comparison of images acquired from the same mouse lung using three imaging techniques (normal breathing / no gating, normal breathing with gating at End Inspiration (EI) and finally the IIBH technique) demonstrated substantial improvements in resolution and quality when using the IIBH gating. Using IIBH triggering the total image acquisition time increased from 15 minutes to 35 minutes, although total x-ray exposure time and hence animal dosage remains the same. This technique is an important step in providing high quality lung imaging of the mouse in vivo, and will provide a good foundation for future longitudinal studies.

Cardiac arrhythmias are a major cause of death (7 million cases annually worldwide; 400,000 in the U.S. alone) and disability. Yet, a noninvasive imaging modality to identify patients at risk, provide accurate diagnosis and guide therapy is not yet available in clinical practice. In my conference presentation and proceedings article, I will describe examples of the application of Electrocardiographic Imaging (ECGI) in humans. ECGI is a new noninvasive imaging modality for cardiac arrhythmias developed in our laboratory. It combines recordings of 224 body-surface electrocardiograms and a thoracic CT scan to reconstruct potentials, electrograms and isochrones (activation sequences) on the heart surface. Examples include: (1) normal activation and repolarization; (2) activation during ventricular pacing; and (3) atrial flutter.

Conventional analysis of cardiac ventricular magnetic resonance images is performed using short axis images and does not guarantee completeness and consistency of the ventricle coverage. In this paper, a four-dimensional (4D, 3D+time) left and right ventricle statistical shape model was generated from the combination of the long axis and short axis images. Iterative mutual intensity registration and interpolation were used to merge the long axis and short axis images into isotropic 4D images and simultaneously correct existing breathing artifact. Distance-based shape interpolation and approximation were used to generate complete ventricle shapes from the long axis and short axis manual segmentations. Landmarks were automatically generated and propagated to 4D data samples using rigid alignment, distance-based merging, and B-spline transform. Principal component analysis (PCA) was used in model creation and analysis. The two strongest modes of the shape model captured the most important shape feature of Tetralogy of Fallot (TOF) patients, right ventricle enlargement. Classification of cardiac images into classes of normal and TOF subjects performed on 3D and 4D models showed 100% classification correctness rates for both normal and TOF subjects using k-Nearest Neighbor (k=1 or 3) classifier and the two strongest shape modes.

The noninvasive assessment of coronary atherosclerosis holds great promise for the future of cardiovascular medicine, and multidetector computed tomography (MDCT) has recently taken the lead in this area. Earlier studies have shown the ability of MDCT to visualize the coronary lumen and various types of atherosclerotic plaque. The aims of this project are to design, implement, and validate a complete system for the automated, quantitative analysis of coronary MDCT images. The developed system uses graph algorithms and knowledge-based cost functions to automatically segment the lumen and wall, and then uses pattern classification techniques to identify and quantify the tissue types found within the detected vascular wall. The system has been validated in comparison with expert tracings and labels, as well as in comparison with intravascular ultrasound (IVUS). In the former, the radial position of the lumen and adventitia were compared at 360 corresponding angular locations in 299 vascular cross sections (from 13 vessels in 5 patients: 5 RCA, 4 LAD, 4 LCX). Results show a border positioning error of 0.150 ± 0.090 mm unsigned / 0.007 ± 0.001 mm signed for the lumen, and 0.210 ± 0.120 mm unsigned / 0.020 ± 0.030 mm signed for the vessel wall. In the comparison with IVUS, the luminal and vascular cross sectional areas were compared in 7 vessels; good correlation was shown for both the lumen (R=0.83) and the vessel wall (R=0.76). The plaque characterization algorithm correctly classified 92% of calcified plaques and 87% of non-calcified plaques.

The adventitia and outer media of large blood vessels are supplied with nutrients by microscopic blood vessels called vasa vasorum. While vasa vasorum have been implicated in a number of diseases including atherosclerosis, knowledge of their functional anatomy and specific role in these diseases has been hindered due to the small size of the vasa vasorum, and difficulty in accessing them. Micro-CT and histological methods have been used in ex-vivo animal studies of the vasa vasorum, but these techniques are limited by their inability to be used for in-vivo investigation. As such, there is very little in-vivo human data available. Intra-vascular ultrasound can acquire high-resolution anatomic images of coronary vessels. ChromaFlo IVUS has been used to identify blood flow in vessel lumens and has exciting prospect for in-vivo studies of vasa vasorum functional anatomy. In this study, ChromaFlo IVUS images of the human mid-left anterior descending coronary artery (LAD) were segmented to analyze the distribution of adventitial vasa vasorum proximal to intimal plaque. Previous animal studies suggest that formation of intimal plaque is accompanied by increased density of adventitial vasa vasorum. The data collected with ChromaFlo ultrasound is inconsistent with the current literature. While IVUS has the fidelity to acquire high-resolution US images of the coronary arteries, ChromaFlo lacks the necessary resolving power to differentiate the vasa vasorum. Further study of IVUS and other imaging methods on a large cohort will provide the basis for future in-vivo analysis of coronary disease.

Recent studies have demonstrated that atherosclerotic plaque surface morphology in the carotid arterial system represents an independent risk factor for embolus formation and subsequent cerebrovascular occlusive events. The primary aim of the current retrospective study is to enhance the clinical utility of this key finding by developing and evaluating objective, quantitative methods for characterizing plaque surface irregularity from Gadolinium-enhanced magnetic resonance angiography (MRA) studies. Nine metrics were analyzed for correlation with percent stenosis in 78 arteries from 43 patients with carotid artery disease. Most of the metrics comprised measurements obtained from a surface model of the stenotic lesion derived from the MRA via the Marching Cubes algorithm with application of the Isosurface Deformable Model. Percent stenosis was determined through real-time volume rendering of 3D MIP MRA studies in Vitrea2. Six of the analyzed metrics revealed significant correlation to percent stenosis (p<0.01). Reproducibility of all metrics was evaluated in a set of 14 randomly selected arteries from 13 patients by way of a single-trial, two-observer analysis. Six of the nine metrics demonstrated significant inter-observer reproducibility by way of single-factor ANOVA analysis (p<0.02). Collectively, the findings reported herein demonstrate an objective and reliable method for quantifying carotid plaque surface irregularity from standard MRA techniques with possible future clinical application in refining risk of ischemic cerebrovascular events and associated need for prophylactic intervention.

Atherosclerosis is characterized by the development of plaques in the arterial wall, which ultimately leads to heart attacks and stroke. 3D ultrasound (US) has been used to screen patients' carotid arteries. Plaque measurements obtained from these images may aid in the management and monitoring of patients, and in evaluating the effect of new treatment options. Different types of measures for ultrasound phenotypes of atherosclerosis have been proposed. Here, we report on the development and application of a method used to analyze changes in carotid plaque morphology from 3D US images obtained at two different time points. We evaluated our technique using manual segmentations of the wall and lumen of the carotid artery from images acquired in two US scanning sessions. To incorporate the effect of intraobserver variability in our evaluation, manual segmentation was performed five times each for the arterial wall and lumen. From this set of five segmentations, the mean wall and lumen surfaces were reconstructed, with the standard deviation at each point mapped onto the surfaces. A correspondence map between the mean wall and lumen surfaces was then established, and the thickness of the atherosclerotic plaque at each point in the vessel was estimated to be the distance between each correspondence pairs. The two-sample Student's t-test was used to judge whether the difference between the thickness values at each pair corresponding points of the arteries in the two 3D US images was statistically significant.

In the past, several methods based on iterative solution of pressure-Poisson equation have been developed for measurement of pressure from phase-contrast magnetic resonance (PC-MR) data. We have developed a novel non-iterative harmonics-based orthogonal projection method which can keep the pressures measured based on the Navier-Stokes equation independent of the path of integration. The gradient of pressure calculated with Navier-Stokes equation is expanded with a series of orthogonal basis functions, and is subsequently projected onto an integrable subspace. Before the projection step however, a scheme is devised to eliminate the discontinuity at the vessel boundaries. The approach was applied to noise-added velocities obtained for both steady and pulsatile stenotic flows from computational fluid dynamics (CFD) simulations and compared with pressures independently obtained by CFD. Additionally, MR velocity data for steady flows measured in in-vitro phantom models with different degree of stenoses and different flow rates were used to test the algorithm and results were compared with CFD simulations.

Finite element method based patient-specific wall stress in abdominal aortic aneurysm (AAA) may provide a more accurate rupture risk predictor than the currently used maximum transverse diameter. In this study, we have investigated the sensitivity of the wall stress in AAA with respect to geometrical variations. We have acquired MR and CT images for four patients with AAA. Three individual users have delineated the AAA vessel wall contours on the image slices. These contours were used to generate synthetic feature images for a deformable model based segmentation method. We investigated the reproducibility and the influence of the user variability on the wall stress. For sufficiently smooth models of the AAA wall, the peak wall stress is reproducible for three out of the four AAA geometries. The 0.99 percentiles of the wall stress show excellent reproducibility for all four AAAs. The variations induced by user variability are larger than the errors caused by the segmentation variability. The influence of the user variability appears to be similar for MR and CT. We conclude that the peak wall stress in AAA is sensitive to small geometrical variations. To increase reproducibility it appears to be best not to allow too much geometrical detail in the simulations. This could be achieved either by using a sufficiently smooth geometry representation or by using a more robust statistical parameter derived from the wall stress distribution.

This study shows the influence of the upstream parent artery geometry on intra-aneurysmal hemodynamics of cerebral aneurysms. Patient-specific models of four cerebral aneurysms at four typical locations were constructed from 3D rotational angiography images. Two geometrical models were constructed for each patient, one with the native parent vessel geometry and another with the parent vessel truncated approximately 1cm upstream from the aneurysm. For one aneurysm, two images were used to construct a model as realistic and large as possible - down to the carotid bifurcation - which was cut at seven different locations. Corresponding finite element grids were generated and computational fluid dynamics simulations were carried out under pulsatile flow conditions. It was found that truncated models tended to underestimate the wall shear stress in the aneurysm and to shift the impaction zone to the neck when compared with the native geometry. In one aneurysm the parent vessel included a tortuous segment close to the neck that strongly influenced the flow pattern entering the aneurysm. Thus, including longer portions of the parent vessel beyond this segment did not have a substantial effect. Depending on the dominant geometrical features the length of the parent artery needed for an accurate representation of the intraaneurysmal hemodynamics may vary among individuals. In conclusion, failure to properly model the inflow stream determined by the upstream parent artery can significantly influence the results of intra-aneurysmal hemodynamic models. The upstream portion of the parent vessel of cerebral aneurysms should be included in order to accurately represent the intraaneurysmal hemodynamics.

One of the factors affecting the accuracy of patient-specific, imaging-based computational hemodynamic studies is the accuracy of geometric models created from medical images. In the present study we have investigated as to how accurate the geometric models should be in the context of cerebral aneurysms in order to obtain an accurate reproduction of intra-aneurysmal hemodynamics in individual patients using numerical simulations. Computed tomography angiography (CTA) images obtained for a patient-specific anterior communicating artery (ACoA) aneurysm and a patient-specific middle cerebral artery (MCA) aneurysm were used to construct the geometric models. For each aneurysm, two models were created, one using a different threshold value for image segmentation than the other. The average distance between the models was about the size of one in-plane pixel. It was found that for the MCA aneurysm, the simulated pressure and shear stress distributions for the two models were entirely different while for the ACoA aneurysm the mean pressure distribution obtained for the two models were similar, but the shear stress distributions were completely different. These results indicate that accurate reproduction of intra-aneurysmal hemodynamics would require the geometric reconstruction from medical images to be highly accurate.

Conventional evaluation of the significance of vascular disease has focused on estimates of geometric factors. There is now substantial interest in investigating whether the onset and progression of vascular pathology can be related to hemodynamic factors. Current imaging modalities have excellent capabilities in delineating the geometric boundaries of the vascular lumen. Advanced non-invasive imaging modalities such as Multi Detector CT and MRI are also able to define the extent of disease within the vessel wall and to provide information on the composition of thrombotic and atherosclerotic components. Finally, it is also possible to use imaging techniques to measure flow velocities across the lumen of vessels of interest, and to determine the pulsatile variation of these velocities through the cardiac cycle. Despite these advanced capabilities, imaging alone is unable to determine important features of the vascular hemodynamics such as wall shear stress or pressure distributions. However, the information on lumenal geometry and the inlet and outlet flow conditions can be used as input into numerical simulation models that are able to predict those quantities. These Computational Fluid Dynamics models can be used to predict hemodynamic parameters on a patient-specific basis. It is therefore possible to use non-invasive imaging methods to follow the progression of vascular disease over time, and to relate changes in lumenal and wall structure to calculated hemodynamic descriptors. This approach can be used not only to understand the natural progression of vascular disease, but as a tool to predict the likely outcome of a surgical intervention.

The accuracy and reproducibility of hemodynamic simulation for a brain aneurysm system was determined by comparison of physical measurements made in a curved duct with the corresponding simulations produced by three different solvers, and by inter-solver comparison of blood flow in a patient-specific, imaging-based model of an aneurysm. The simulations were in close agreement with measurements made in the square duct. This suggests that hemodynamic simulation is accurate for models with strong curvature flow. The simulation results produced by solvers using the model of the brain aneurysm were consistent with each other, suggesting that hemodynamic simulations of patient-specific imaging-based aneurysm models are consistent and reproducible by different solvers. These results support the validity of patient-specific imaging-based simulations.

In this study, the effects of unequal physiologic flow conditions in the internal carotid arteries on the intra-aneurysmal hemodynamics of anterior communicating artery aneurysms were investigated. Patient-specific vascular computational fluid dynamics models of five cerebral aneurysms were constructed from bilateral 3D rotational angiography images. The aneurysmal hemodynamics was analyzed under a range of physiologic flow conditions including the effects of unequal mean flows and phase shifts between the flow waveforms of the left and right internal carotid arteries. A total of five simulations were performed for each patient, and unsteady wall shear stress (WSS) maps were created for each flow condition. Time dependent curves of average WSS magnitude over selected regions on the aneurysms were constructed and used to analyze the influence of the inflow conditions. It was found that mean flow imbalances in the feeding vessels tend to shift the regions of elevated WSS (flow impingement region) towards the dominating inflow jet and to change the magnitude of the WSS peaks. However, the overall qualitative appearance of the WSS distribution and velocity simulations is not substantially affected. In contrast, phase differences tend to increase the temporal complexity of the hemodynamic patterns and to destabilize the intra-aneurysmal flow pattern. However, these effects are less important when the A1 confluence is less symmetric, i.e. dominated by one of the A1 segments. Conditions affecting the flow characteristics in the parent arteries of cerebral aneurysms with more than one avenue of inflow should be incorporated into flow models.

An asymmetric stent with low porosity patch across the intracranial aneurysm neck and high porosity elsewhere is designed to modify the flow to result in thrombogenesis and occlusion of the aneurysm and yet to reduce the possibility of also occluding adjacent perforator vessels. The purposes of this study are to evaluate the flow field induced by an asymmetric stent using both numerical and digital subtraction angiography (DSA) methods and to quantify the flow dynamics of an asymmetric stent in an in vivo aneurysm model. We created a vein-pouch aneurysm model on the canine carotid artery. An asymmetric stent was implanted at the aneurysm, with 25% porosity across the aneurysm neck and 80% porosity elsewhere. The aneurysm geometry, before and after stent implantation, was acquired using cone beam CT and reconstructed for computational fluid dynamics (CFD) analysis. Both steady-state and pulsatile flow conditions using the measured waveforms from the aneurysm model were studied. To reduce computational costs, we modeled the asymmetric stent effect by specifying a pressure drop over the layer across the aneurysm orifice where the low porosity patch was located. From the CFD results, we found the asymmetric stent reduced the inflow into the aneurysm by 51%, and appeared to create a stasis-like environment which favors thrombus formation. The DSA sequences also showed substantial flow reduction into the aneurysm. Asymmetric stents may be a viable image guided intervention for treating intracranial aneurysms with desired flow modification features.

Modeling aneurysm growth using stress-mediated growth laws requires accurate knowledge of the hemodynamic stresses and strains acting on the aneurysm wall due to the internal blood flow and the external tissue support. Therefore, solving the coupled problem of blood flow and vessel wall deformation represents a critical step in the evaluation of these hemodynamic stresses, but for large, patient-specific models of the vasculature one that is computationally expensive. In this work, we present the application of a new formulation, the Coupled Momentum Method for Fluid-Solid Interaction (CMM-FSI), to compute blood flow and vessel wall deformation under realistic ranges of pressures for large patient-specific models of the cerebro-vasculature. The method couples the equations of the deformation of the vessel wall at the variational level as a boundary condition for the fluid domain. We consider a strong coupling of the degrees-of-freedom of the fluid interface and the wall domains. The effect of the vessel wall boundary is therefore added in a monolithic way to the fluid equations, resulting in a remarkably robust and computationally-efficient scheme. The method is applied to patient-specific model of the Circle of Willis featuring a saccular aneurysm, using resistance outflow boundary conditions. The wall normal and shear stresses resulting from the simulation can then be used as the hemodynamic forces mediating the aneurysm wall adaptation in the algorithm shown in the second part of this work.

Computational Fluid Dynamics (CFD) techniques were used to assess the dynamic pressure distribution at the sites of three virtually removed paraclinoid aneurysms. 3D Digital Subtraction Angiograms (3D-DSA) were used to create two computational meshes for each aneurysm. The first mesh was derived directly from the angiographic data containing the aneurysms. In the second mesh, the section of the artery across the aneurysm neck was virtually reconstructed by interpolation between the adjacent proximal and distal parts of the parent artery. This mesh is considered to be an approximation to the geometry of the parent artery before aneurysm formation. Results from the simulations showed that the peak dynamic pressures occurred at arterial bends. The change in the magnitude of dynamic pressure over the cycle was about 7.5 mmHg which is about 15% of the change in the background static pressure. However, further study is needed to clarify if this change in dynamic pressure would elicit any biochemical response from the vessel tissue.

In this study we utilize our texture characterization software (3-D AMFM) to characterize interstitial lung diseases (including emphysema) based on MDCT generated volumetric data using 3-dimensional texture features. We have sought to test whether the scanner and reconstruction filter (kernel) type affect the classification of lung diseases using the 3-D AMFM. We collected MDCT images in three subject groups: emphysema (n=9), interstitial pulmonary fibrosis (IPF) (n=10), and normal non-smokers (n=9). In each group, images were scanned either on a Siemens Sensation 16 or 64-slice scanner, (B50f or B30 recon. kernel) or a Philips 4-slice scanner (B recon. kernel). A total of 1516 volumes of interest (VOIs; 21x21 pixels in plane) were marked by two chest imaging experts using the Iowa Pulmonary Analysis Software Suite (PASS). We calculated 24 volumetric features. Bayesian methods were used for classification. Images from different scanners/kernels were combined in all possible combinations to test how robust the tissue classification was relative to the differences in image characteristics. We used 10-fold cross validation for testing the result. Sensitivity, specificity and accuracy were calculated. One-way Analysis of Variances (ANOVA) was used to compare the classification result between the various combinations of scanner and reconstruction kernel types. This study yielded a sensitivity of 94%, 91%, 97%, and 93% for emphysema, ground-glass, honeycombing, and normal non-smoker patterns respectively using a mixture of all three subject groups. The specificity for these characterizations was 97%, 99%, 99%, and 98%, respectively. The F test result of ANOVA shows there is no significant difference (p <0.05) between different combinations of data with respect to scanner and convolution kernel type. Since different MDCT and reconstruction kernel types did not show significant differences in regards to the classification result, this study suggests that the 3-D AMFM can be generally introduced.

Tree matching methods have numerous applications in medical imaging, including registration, anatomical labeling, segmentation, and navigation of structures such as vessels and airway trees. Typical methods for tree matching rely on conventional graph matching techniques and therefore suffer potential limitations such as sensitivity to the accuracy of the extracted tree structures, as well as dependence on the initial alignment. We present a novel path-based tree matching framework independent of graph matching. It is based on a point-by-point feature comparison of complete paths rather than branch points, and consequently is relatively unaffected by spurious airways and/or missing branches. A matching matrix is used to enforce one-to-one matching. Moreover our method can reliably match irregular tree structures, resulting from imperfect segmentation and centerline extraction. Also reflecting the nature of these features, our method does not require a precise alignment or registration of tree structures. To test our method we used two thoracic CT scans from each of ten patients, with a median inter-scan interval of 3 months (range 0.5 to 10 months). The bronchial tree structure was automatically extracted from each scan and a ground truth of matching paths was established between each pair of tree structures. Overall 87% of 702 airway paths (average 35.1 per patient matched both ways) were correctly matched using this technique. Based on this success we also present preliminary results of airway-to-artery matching using our proposed methodology.

When addressing the task of automatic tracheo-bronchial tree matching it seems natural to make use of the tree's graph structure and branching characteristics. In tracheo-bronchial trees that were automatically extracted from multi-slice CT data, however, the graph information is not always reliable, especially for noisy or low-dose data which makes the abovementioned class of approaches prone to error in these situations. In this work we investigate what can be gained by using the spatial position of the bronchial centerline points. For this purpose we introduce, investigate, and compare two approaches to tree matching that are based on the use of centerline point positions alone with no additional connectivity information. As features we use (1) the 3D shape context and (2) statistical moments of the local point distribution. The 3D shape context has recently been introduced as a regional shape descriptor. It is based on a spherical histogram with logarithmic sampling in the radial direction. Both methods are used in order to match an automatically extracted tree to a manually labeled model tree which results in an automatic anatomical labeling of the data tree. Six tracheo-bronchial trees were matched to a given model tree. The data trees covered a range from high quality data to poor quality data. Furthermore two of the cases exhibited strongly distorted anatomy. It could be shown that the 3D shape context feature labeled 69 % of the branches correctly with one of 34 anatomical labels. In the case of the statistical moment feature 40 % of the branches were labeled correctly. We conclude that the set of centerline points alone allows correct labeling of a large portion of lung segments. We propose to combine the valuable local shape information in future work with connectivity and branching information, where the latter is reliably available.

Respiration causes movement and shape changes in thoracic tumors, which has a direct influence on the radio-therapy planning process. Current methods for the estimation of tumor mobility are either two-dimensional (fluoroscopy, 2D dynamic MRI) or based on radiation (3D (+t) CT, implanted gold markers). With current advances in dynamic MRI acquisition, 3D+t image sequences of the thorax can be acquired covering the thorax over the whole breathing cycle. In this work, methods are presented for the interactive segmentation of tumors in dynamic images, the calculation of tumor trajectories, dynamic tumor volumetry and dynamic tumor rotation/deformation based on 3D dynamic MRI. For volumetry calculation, a set of 21 related partial volume correcting volumetry algorithms has been evaluated based on tumor surrogates. Conventional volumetry based on voxel counting yielded a root mean square error of 29% compared to a root mean square error of 11% achieved by the algorithm performing best among the different volumetry methods. The new workflow has been applied to a set of 26 patients. Preliminary results indicate, that 3D dynamic MRI reveals important aspects of tumor behavior during the breathing cycle. This might imply the possibility to further improve high-precision radiotherapy techniques.

Pulmonary diseases such as bronchiectasis, asthma, and emphysema are characterized by abnormalities in airway dimensions. Multi-slice computed tomography (MSCT) has become one of the primary means to depict these abnormalities, as the availability of high-resolution near-isotropic data makes it possible to evaluate airways at oblique angles to the scanner plane. However, currently, clinical evaluation of airways is typically limited to subjective visual inspection only: systematic evaluation of the airways to take advantage of high-resolution data has not proved practical without automation. We present an automated method to quantitatively evaluate airway lumen diameter, wall thickness and broncho-arterial ratios. In addition, our method provides 3D visualization of these values, graphically illustrating the location and extent of disease. Our algorithm begins by automatic airway segmentation to extract paths to the distal airways, and to create a map of airway diameters. Normally, airway diameters decrease as paths progress distally; failure to taper indicates abnormal dilatation. Our approach monitors airway lumen diameters along each airway path in order to detect abnormal profiles, allowing even subtle degrees of pathologic dilatation to be identified. Our method also systematically computes the broncho-arterial ratio at every terminal branch of the tree model, as a ratio above 1 indicates potentially abnormal bronchial dilatation. Finally, the airway wall thickness is computed at corresponding locations. These measurements are used to highlight abnormal branches for closer inspection, and can be summed to compute a quantitative global score for the entire airway tree, allowing reproducible longitudinal assessment of disease severity. Preliminary tests on patients diagnosed with bronchiectasis demonstrated rapid identification of lack of tapering, which also was confirmed by corresponding demonstration of elevated broncho-arterial ratios.

We introduce an automatic 3D multiscale automatic segmentation algorithm for delineating specific organs in Magnetic Resonance images (MRI). The algorithm can process several modalities simultaneously, and handle both isotropic and anisotropic data in only linear time complexity. It produces a hierarchical decomposition of MRI scans. During this segmentation process a rich set of features describing the segments in terms of intensity, shape and location are calculated, reflecting the formation of the hierarchical decomposition. We show that this method can delineate the entire uterus of the rat abdomen in 3D MR images utilizing a combination of scanning protocols that jointly achieve high contrast between the uterus and other abdominal organs and between inner structures of the rat uterus. Both single and multi-channel automatic segmentation demonstrate high correlation to a manual segmentation. While the focus here is on the rat uterus, the general approach can be applied to recognition in 2D, 3D and multi-channel medical images.

Delineation of objects within medical images is often difficult to perform reproducibly when one relies upon hand-segmentation. To avoid inter- and intra-user variability, a semi-automatic segmentation method can more accurately and consistently determine the object boundaries. This paper presents a semi-automatic process for determining the length and volume of the spinal cord between adjacent pairs of intervertebral discs and the total length and volume of the spinal cord. A level set segmentation was performed on MRI data with user selected landmarks in order to obtain a segmentation of the spinal cord. The length and volume measurements were performed on 20 segments from C1 to L1 with five sets of user selected landmarks. Our results show that the average spinal cord segment length was 21.55 mm with a standard deviation of 25.11% and the average spinal cord segment volume was 2,217.16 mm³ with a standard deviation of 80.51%. The measurement variability of a single anatomical length across multiple trials of different sets of seed points was three orders of magnitude lower (0.06%) than the variability across different anatomical lengths (25.23%), while the measurement variability of a single anatomical volume across multiple trials of different sets of seed points was two orders of magnitude lower (0.37%) than the variability across different anatomical volumes (79.24%). Our method has been demonstrated to be potentially insensitive to intra- and inter-user variability.

Magnetic resonance (MR) imaging is an imaging modality that is used in the management and diagnosis of acute stroke. Common MR imaging techniques such as diffusion weighted imaging (DWI) and apparent diffusion coefficient maps (ADC) are used routinely in the diagnosis of acute infarcts. However, advances in radiology information systems and imaging protocols have led to an overload of image information that can be difficult to manage and time consuming. Automated techniques to assist in the identification of acute ischemic stroke can prove beneficial to 1) the physician by providing a mechanism for early detection and 2) the patient by providing effective stroke therapy at an early stage. We have processed DW images and ADC maps using a novel automated Relative Difference Map (RDM) method that was tailored to the identification and delineation of the stroke region. Results indicate that the technique can delineate regions of acute infarctions on DW images and ADC maps. A formal evaluation of the RDM algorithm was performed by comparing accuracy measurements between 1) expert generated ground truths with the RDM delineated DWI infarcts and 2) RDM delineated DWI infarcts with RDM delineated ADC infarcts. The accuracy measurements indicate that the RDM delineated DWI infarcts are comparable to the expert generated ground truths. The true positive volume fraction value (TPVF), between RDM delineated DWI and ADC infarcts, is nonzero for all cases with an acute infarct while the value for non-acute cases remains zero.

Multiple Sclerosis (MS) is an inflammatory and demyelinating disorder of the central nervous system with a presumed immune-mediated etiology. For treatment of MS, the measurements of white matter (WM), gray matter (GM), and cerebral spinal fluid (CSF) are often used in conjunction with clinical evaluation to provide a more objective measure of MS burden. In this paper, we apply a new unifying automatic mixture-based algorithm for segmentation of brain tissues to quantitatively analyze MS. The method takes into account the following effects that commonly appear in MR imaging: 1) The MR data is modeled as a stochastic process with an inherent inhomogeneity effect of smoothly varying intensity; 2) A new partial volume (PV) model is built in establishing the maximum a posterior (MAP) segmentation scheme; 3) Noise artifacts are minimized by a priori Markov random field (MRF) penalty indicating neighborhood correlation from tissue mixture. The volumes of brain tissues (WM, GM) and CSF are extracted from the mixture-based segmentation. Experimental results of feasibility studies on quantitative analysis of MS are presented.

This paper presents (1) an improved hierarchical method for segmenting the component tissue regions in fast spin echo T2 and PD images of the brain of Multiple Sclerosis (MS) patients, and (2) a methodology to characterize the disease utilizing the distributions of standardized T2 and PD intensities in the segmented tissue regions. First, the background intensity inhomogeneities are corrected and the intensity scales are standardized for all acquired images. The segmentation method imposes a feedback-like procedure on our previously developed hierarchical brain tissue segmentation method. With gradually simplified patterns in images and stronger evidences, pathological objects are recognized and segmented in an interplay fashion. After the brain parenchymal (BP) mask is generated, an under-estimated gray matter mask (uGM) and an over-estimated white matter mask (oWM) are created. Pure WM (PWM) and lesion (LS) masks are extracted from the all-inclusive oWM mask. By feedback, accurate GM and WM masks are subsequently formed. Finally, partial volume regions of GM and WM as well as Dirty WM (DWM) masks are generated. Intensity histograms and their parameters (peak height, peak location, and 25th, 50th and 75th percentile values) are computed for both T2 and PD images within each tissue region. Tissue volumes are also estimated. Spearman correlation coefficient rank test is then utilized to assess if there exists a trend between clinical states and the image-based parameters. This image analysis method has been applied to a data set consisting of 60 patients with MS and 20 normal controls. LS related parameters and clinical Extended Disability Status Scale (EDSS) scores demonstrate modest correlations. Almost every intensity-based parameter shows statistical difference between normal control and patient groups with a level better than 5%. These results can be utilized to monitor disease progression in MS.

Osteoporosis is the cause of over 1.5 million bone fractures annually. Most of these fractures occur in sites rich in trabecular bone, a complex network of bony struts and plates found throughout the skeleton. The three-dimensional structure of the trabecular bone network significantly determines mechanical strength and thus fracture resistance. Here we present a data acquisition and processing system that allows efficient noninvasive assessment of trabecular bone structure through a "virtual bone biopsy". High-resolution MR images are acquired from which the trabecular bone network is extracted by estimating the partial bone occupancy of each voxel. A heuristic voxel subdivision increases the effective resolution of the bone volume fraction map and serves a basis for subsequent analysis of topological and orientational parameters. Semi-automated registration and segmentation ensure selection of the same anatomical location in subjects imaged at different time points during treatment. It is shown with excerpts from an ongoing clinical study of early post-menopausal women, that significant reduction in network connectivity occurs in the control group while the structural integrity is maintained in the hormone replacement group. The system described should be suited for large-scale studies designed to evaluate the efficacy of therapeutic intervention in subjects with metabolic bone disease.

Biomechanical testing is the gold standard to determine bone competence, and has been used extensively. Direct mechanical testing provides detailed information on overall bone mechanical and material properties, but fails in revealing local properties such as local deformations and strains or quantification of fracture progression. Therefore, we incorporated several imaging methods in our mechanical setups in order to get a better insight into bone deformation and failure characteristics. Our aim was to develop an integrative approach for hierarchical investigation of bone, working at different scales of resolution ranging from the whole bone to its ultrastructure. At a macroscopic level, we used high-resolution and high-speed cameras which drastically increased the amount of information obtained from a biomechanical bone test. The new image data proved especially important when dealing with very small bones such as the murine femur. Here the feedback of the camera in the process of aligning and positioning the samples is indispensable for reproducibility. In addition, global failure behavior and fracture initiation can now be visualized with high temporal resolution. At a microscopic level, bone microstructure, i.e. trabecular architecture and cortical porosity, are known to influence bone strength and failure mechanisms significantly. For this reason, we developed an image-guided failure assessment technique, also referred to as functional microimaging, allowing direct time-lapsed 3D visualization and computation of local displacements and strains for better quantification of fracture initiation and progression at the microscopic level. While the resolution of typical desktop micro-computed tomography is around a few micrometers, highly brilliant X-rays from synchrotron radiation permit to explore the nanometer world. This allowed, for the first time, to uncover fully nondestructively the 3D ultrastructure of bone including vascular and cellular structures and to investigate their role in development of bone microcracks in an unprecedented resolution. We conclude that functional microimaging, i.e. the combination of biomechanical testing with non-destructive 3D imaging and visualization are extremely valuable in studying bone failure mechanisms. Functional investigation of microcrack initiation and propagation will lead to a better understanding of the relative contribution of bone mass and bone quality to bone competence.

MR elastography (MRE) images the intrinsic mechanical properties of soft tissues; e.g., the shear modulus, μ. The μ of the plantar soft tissues is important in understanding the mechanisms whereby the forces induced during normal motion produce ulcers that lead to amputation in diabetic feet. We compared the compliance of the heel fat pad to compressive forces and to shearing forces. The design of prosthetics to protect the foot depends on the proper understanding of the mechanisms inducing damage. In the heel fat pads of six normal subjects, between 25 and 65 years of age, the μ for deformation perpendicular to the direction of weight bearing is similar but not identical to that determined for deformation along the weight bearing axis. The average difference between μ along the weight bearing axis and μ perpendicular to the weight bearing axis, is well correlated with age (Correlation Coefficient = 0.789). The p-value for the data being random was 0.0347 indicating that the observed difference is not likely to be random. The p-value for control points is 0.8989, indicating a random process. The results are suggestive that the high compressive forces imposed during walking damage the heel fat pads over time resulting in softening to compression preferentially over shearing. It is important to validate the observed effect with larger numbers of subjects, and better controls including measures of activity, and to understand if diseases like diabetes increase the observed damage.

PURPOSE: To assess the feasibility of a modified phase-contrast MRI technique (MR Elastography) for quantitatively assessing the mechanical properties of hepatic tissues by imaging propagating acoustic shear waves. MATERIALS AND METHODS: Both phantom and human studies were performed to develop and optimize a practical imaging protocol by visualizing and investigating the diffraction field of shear waves generated from pneumatic longitudinal drivers. The effects of interposed ribs in a transcostal approach were also investigated. A gradient echo MRE pulse sequence was adapted for shear wave imaging in the liver during suspended respiration, and then tested to measure hepatic shear stiffness in 13 healthy volunteers and 1 patient with chronic liver disease to determine the potential of non-invasively detecting liver fibrosis. RESULTS: Phantom studies demonstrate that longitudinal waves generated by the driver are mode-converted to shear waves in a distribution governed by diffraction principles. The transcostal approach was determined to be the most effective method for generating shear waves in human studies. Hepatic stiffness measurements in the 13 normal volunteers demonstrated a mean value of 2.0±0.2kPa. The shear stiffness measurement in the patient was much higher at 8.5kPa. CONCLUSION: MR Elastography of the liver shows promise as a method to non-invasively detect and characterize diffuse liver disease, potentially reducing the need for biopsy to diagnose hepatic fibrosis.

This work extends a recently realized inverse problem technique of extracting soft tissue elasticity information via non-rigid model-based image registration. The algorithm uses the elastic properties of the tissue in a biomechanical model to achieve maximal similarity between image data acquired under different states of loading. A new multi-resolution, non-linear optimization framework has been employed which allows for improved performance and object detection. Prior studies have demonstrated successful reconstructions from images of a tissue-like thin membrane phantom with a single embedded inclusion that was significantly stiffer than its surroundings. For this investigation, a similar phantom was fabricated with two stiff inclusions to test the effectiveness of this method in discriminating multiple smaller objects. Elasticity values generated from both simulation and real data testing scenarios provided sufficient contrast for detection and good quantitative localization of the inclusion areas.

Proper targeting of radiation therapy during the treatment of prostate cancer requires the successful alignment of initial planning images with more recent ones taken during the treatment course. Prostatic displacement, if unaccounted for during the treatment course, can lead to radiation underdosage of the target area or radiation of the surrounding healthy tissue. Studies have shown that prostatic displacement is directly correlated with changes in the volume of the rectum. We have developed a non-rigid image registration system based on a biomechanical model of the prostate, rectum, and surrounding tissues, that incorporates statistical shape information about changes in the volume of the rectum. Using finite-element analysis and statistical shape models, our non-rigid registration method defines a mapping between two prostate image volumes. The proposed method assumes that the prostate image misregistration occurs as a result of changes in the rectum's shape. The change along the rectum's circumference was considered as the displacement boundary condition of the prostate's finite element model. As such we used a mutual information similarity measure in conjunction with the finite element model for computing the optimal boundary condition as well as estimating the location and relative Young's modulus of the central and peripheral zones within the prostate to the surrounding tissue. Compared to other techniques, this registration technique is not only efficient but also capable of providing valuable mechanical properties of tissue in vivo.

Magnetic resonance imaging (MRI) with an endorectal receiver coil (ERC) provides superior visualization of the prostate gland and its surrounding anatomy at the expense of large anatomical deformation. The ability to correct for this deformation is critical to integrate the MR images into the CT-based treatment planning for radiotherapy. The ability to quantify and understand the physiological motion due to large changes in rectal filling can also improve the precision of image-guided procedures. The purpose of this study was to understand the biomechanical relationship between the prostate, rectum, and bladder using a finite element-based multi-organ deformable image registration method, 'Morfeus' developed at our institution. Patients diagnosed with prostate cancer were enrolled in the study. Gold seed markers were implanted in the prostate and MR scans performed with the ERC in place and its surrounding balloon inflated to varying volumes (0-100cc). The prostate, bladder, and rectum were then delineated, converted into finite element models, and assigned appropriate material properties. Morfeus was used to assign surface interfaces between the adjacent organs and deform the bladder and rectum from one position to another, obtaining the position of the prostate through finite element analysis. This approach achieves sub-voxel accuracy of image co-registration in the context of a large ERC deformation, while providing a biomechanical understanding of the multi-organ physiological relationship between the prostate, bladder, and rectum. The development of a deformable registration strategy is essential to integrate the superior information offered in MR images into the treatment planning process.

Previous research has shown that CT-image-based guidance could be useful for the bronchoscopic assessment of lung cancer. This research drew upon the registration of bronchoscopic video images to CT-based endoluminal renderings of the airway tree. The proposed methods either were restricted to discrete single-frame registration, which took several seconds to complete, or required non-real-time buffering and processing of video sequences. We have devised a fast 2D/3D image registration method that performs single-frame CT-Video registration in under 1/15th of a second. This allows the method to be used for real-time registration at full video frame rates without significantly altering the physician's behavior. The method achieves its speed through a gradient-based optimization method that allows most of the computation to be performed off-line. During live registration, the optimization iteratively steps toward the locally optimal viewpoint at which a CT-based endoluminal view is most similar to a current bronchoscopic video frame. After an initial registration to begin the process (generally done in the trachea for bronchoscopy), subsequent registrations are performed in real-time on each incoming video frame. As each new bronchoscopic video frame becomes available, the current optimization is initialized using the previous frame's optimization result, allowing continuous guidance to proceed without manual re-initialization. Tests were performed using both synthetic and pre-recorded bronchoscopic video. The results show that the method is robust to initialization errors, that registration accuracy is high, and that continuous registration can proceed on real-time video at >15 frames per sec. with minimal user-intervention.

This paper presents a method for identifying branches for CT-guided bronchoscopy based on eigenspace image matching. This method outputs the current location of a real bronchoscope (RB) by displaying branches where a bronchoscope is currently observing or by presenting anatomical names of branches currently being observed. In the previous method of bronchoscope navigation, the motion of a real bronchoscope is tracked by image registration between RB and virtual bronchoscopic (VB) images. Although bronchoscope tracking based on image registration gives us very accurate tracking results, it requires a lot of computation time and it is difficult to perform real-time tracking. If we focus only on navigation to a target branch, it is enough to identify a branch where a bronchoscope is currently located. This paper presents a method for identifying branches in which a bronchoscope is currently observing and presenting their anatomical names. Branch identification is done by image matching between RB images and pre-generated VB images. VB images are pre-generated at each bifurcation point based on structural analysis results of bronchi regions extracted from CT images. For each frame of an RB video, we find the most similar VB image to the input one from a training dataset (pre-generated VB image) and output the branch levels associated with the found image by using the eigenspace method. We have applied the proposed method to a pair of comprising a 3D CT image and real bronchoscopic video footage. The experimental results showed that the proposed method can identify branches for about 77.7% of the input frames.

We present a synchronous navigation module for CT colonography (CTC) reading. The need for such a system arises because most CTC protocols require a patient to be scanned in both supine and prone positions to increase sensitivity in detecting colonic polyps. However, existing clinical practices are limited to reading one scan at a time. Such limitation is due to the fact that building a reference system between scans for the highly flexible colon is a nontrivial task. The conventional centerline approach, generating only the longitudinal distance along the colon, falls short in providing the necessary orientation information to synchronize the virtual navigation cameras in both scanned positions. In this paper we describe a synchronous navigation system by using the teniae coli as anatomical references. Teniae coli are three parallel bands of longitudinal smooth muscle on the surface of the colon. They are morphologically distinguishable and form a piecewise triple helix structure from the appendix to the sigmoid colon. Because of these characteristics, they are ideal references to synchronize virtual cameras in both scanned positions. Our new navigation system consists of two side-by-side virtual colonoscopic view panels (for the supine and prone data sets respectively) and one single camera control unit (which controls both the supine and prone virtual cameras). The capability to examine the same colonic region simultaneously in both scanned images can raise an observer's confidence in polyp identification and potentially improve the performance of CT colonography.

Cardiac arrhythmias are a debilitating, potentially life threatening condition involving aberrant electrical activity in the heart which results in abnormal heart rhythm. Virtual cardioscopy can play an important role in minimally invasive treatment of cardiac arrhythmias. Second and third generation image-guidance systems are now available for the treatment of arrhythmias using RF ablation catheters. While these 3D tools provide useful information to the clinician, additional enhancements to the virtual cardioscopy display paradigm are critical for optimal therapy guidance. Based on input from clinical collaborators, several key visualization techniques have been developed to enhance the use of virtual cardioscopy during cardiac ablation procedures. We have identified, designed and incorporated several visual cues important to successful virtual cardioscopy. These features include the use of global reference maps, parametric mapping, and focused navigation and targeting using abnormal electro-physiologic activity. Our virtual cardioscopy system is designed for real-time use during RF cardiac ablation procedures. Several unique visualizations from our virtual cardioscopy system will be presented. Evaluation of the system with phantom and animal studies will be presented. This research is supported by grant EB002834 from the National Institute of Biomedical Imaging and Bioengineering, National Institutes of Health.

This paper presents an improved method for virtually unfolding an organ and visualizing its entire luminal surface in only one view. Unfolded tract views can be very useful as they allow doctors to understand various kinds of information of the luminal surface intuitively, just like observing a pathological specimen. However, the previous method cannot correctly reproduce the luminal surface because elasticity for the organ walls is quite coarse defined. Thus, three improvements are proposed: (1) accurate elastic modeling using mass points and Kelvin-Voigt visco-elastic elements, (2) stable image deformation by the Newmark-β method, and (3) automatically directing organ walls to flat shapes by forces determined from their surface normals. Unfolded views generated by the proposed method from seventeen 3D CT image datasets are compared with those by the previous method, virtual endoscopic images, and pathological specimens. Several regions on the luminal surface, which could not be reproduced by the previous method, were accurately reproduced. Bending and concave parts of organ walls, which were difficult to unfold in the previous method, were satisfactorily flattened by introducing improved deformation processes. Computation time was reasonably reduced. Unfolded views of twelve cases were presented to doctors for surgical planning. The unfolded views generated by the proposed method were considered to have well reproduced all lesions as well as fold patterns observed in virtual endoscopic images.

We have evaluated the feasibility of polyp detection on simulated ultra low dose CT Colonography data by a computer aided polyp detection (CAD) algorithm. We compared the results of ultra low dose to normal dose data. Twenty-three extensively prepared patients were scanned in prone and supine position at 25 to 100 mAs (average 70 mAs) depending on their waist circumference. Noise was added and the scans were reconstructed at 6.25 and 1.39 mAs. To evaluate the performance of the CAD system, polyps detected by an experienced reviewer and confirmed at colonoscopy were used as ground truth. Curvature, concavity and sphericity of the colon surface were used to detect polyp candidates. Bilateral filtering was used to reduce noise. We present the results for 40 polyps of 6 mm or larger as measured during colonoscopy. The by-polyp sensitivity was 80% for medium size polyps (6-9 mm) and 97% for large polyps (10 mm or larger) at an average value of 5 false-positives per scan for normal dose data. The by-polyp sensitivity was 81% for medium size polyps and 85% for large size polyps at an average value of 5 false-positives per scan for low dose data (6.25 mAs). Finally for the ultra low dose data (1.39 mAs) we achieved a by-polyp sensitivity of 75% for medium size polyps and 97% for large polyps at an average value of 5 false-positives per scan. The conclusion of our study is that CAD for polyp detection is feasible on ultra low dose CT colonography data.

Performance of Computed Tomographic Colonography (CTC) Computer Aided Detection (CTC CAD) depends sensitively on a set of features chosen to characterize a polyp candidate. Most of the features are derived from some shape related characteristics which are calculated at the points located within the boundaries of a polyp candidate. This approach ignores information from the part of the colonic wall stretching beyond the limits of a polyp candidate. We found that almost 90% of small and medium size polyps missed by our CAD program were located on haustral folds. This suggests that two different classifiers (for detections on a fold and not on a fold) could be used which are better tuned to the local characteristics. Therefore, we developed an automated method to verify independently if a given polyp candidate is located on a fold. This is done by checking the intensity profile along normals originating from the points on the colonic wall which are close to but outside of a patch marking a polyp candidate.

We introduce an intuitive measure of computer aided detection (CAD) system performance that can handle simultaneous variation of multiple parameters. On the example of CAD of colon polyps we demonstrate how this measure was used to find the optimal parameters and make improvements to the "water-plane" algorithm that finds initial polyp candidates on the colon wall. In particular, we improved the merging of overlapping clusters to only create fused clusters if they shared at least 50% of their vertices and adjusted size and thickness filter criteria to retain more true positive detections. The system, containing all optimizations, improved significantly over the original system that found initial detections based only on colon surface curvature. This improvement was measured by both free response operating curve (FROC) analysis and our new performance measure.

Virtual colonoscopy is becoming a more prevalent way to diagnose colon cancer. One of the critical elements in detecting cancerous polyps using virtual colonoscopy, especially in conjunction with computer-aided detection of polyps, is that the colon be sufficiently distended. We have developed an automatic method to determine from a CT scan what percentage of the colon is distended by 1cm or larger and compared our method with a radiologist's assessment of quality of the scan with respect to successful colon polyp detection. A radiologist grouped 41 CT virtual colonoscopy scans into three groups according to the degree of colonic distention, "well", "medium", and "poor". We also employed a subvoxel accurate centerline algorithm and a subvoxel accurate distance transform to each dataset to measure the colon distention along the centerline. To summarize the colonic distention with a single value relevant for polyp detection, the distention score, we recorded the percentage of centerline positions in which the colon distention was 1cm or larger. We then compared the radiologist's assessment and the computed results. The sorting of all datasets according to the distention score agreed with the radiologist's assessment. The "poor" cases had a mean and standard deviation score of 78.4% ± 5.2%, the "medium" cases measured 88.7% ± 1.9%, and the "well" cases 98.8% ± 1.5%. All categories were shown to be significantly different from each other using unpaired two sample t-tests. The presented colonic distention score is an accurate method for assessing the quality of colonic distention for CT colonography.

Recently, virtual colonoscopy (VC) has received attention as a new colon diagnostic method based on the development of imaging devices. VC is considered a less invasive inspection and reduces diagnosing time. However, because the colon has many folds and its shape is long and convoluted, a physician has to repeatedly change viewpoints and viewing directions many times. We previously proposed a new computer aided diagnosis system for the colon that provided virtual unfolded (VU) and virtual endoscopic (VE) views of the colon. This system enables physicians to observe a large area of the colonic wall on a VU view synchronized with a VE view. Thus a suspicious area on a VU view can be observed in more detail on a VE view. We generated VU views by controlling the ray directions of volume rendering. This method had a problem: rays intersected at the curved areas of the colon because the rays were cast perpendicular to the central path of the colon. Ray intersections caused spurious holes in the VU views. In this paper, we present a method that reduces ray intersections by employing a model that allocates springs between planes perpendicular to the central path. Then plane directions are modified by the spring forces to minimize the total length of the springs. In a drawing process, a VU view is generated by casting rays along the planes. We applied the method to abdominal CT images. Experimental results showed that the method can reduce spurious holes on VU views more than the previous method.

The rectum can distend to accommodate stool, and contracts in response to distention during defecation. Rectal motor dysfunctions are implicated in the pathophysiology of functional defecation disorders and fecal incontinence. These rectal motor functions can be studied by intra-luminal measurements of pressure by manometry, or combined with volume during rectal balloon distention. Pressure-volume (p-v) relationships provide a global index of rectal mechanical properties. However, balloon distention alone does not measure luminal radius or wall thickness, which are necessary to compute wall tension and stress respectively. It has been suggested that the elastic modulus, which is the linear slope of the stress-strain relationship, is a more accurate measure of wall stiffness. Also, measurements of compliance may not reflect differences in rectal diameter between subjects prior to inflation, and imaging is necessary to determine if, as has been suggested, rectal pressure-volume relationships are affected by extra-rectal structures. We have developed a technique to measure rectal stress:strain relationships in humans, by simultaneous magnetic resonance imaging (MRI) during rectal balloon distention. After a conditioning distention, a rectal balloon was distended with water from 0 to 400 ml in 50 ml steps, and imaged at each step with MRI. The fluid filled balloon was segmented from each volume, the phase-ordered binary volumes were transformed into a geometric characterization of the inflated rectal surface. Taken together with measurements of balloon pressure and of rectal wall thickness, this model of the rectal surface was used to calculate regional values of curvature, tension, strain, and stress for the rectum. In summary, this technique has the unique ability to non-invasively measure the rectal stress:strain relationship and also determine if rectal expansion is limited by extra-rectal structures. This functional information allows the direct clinical analysis of rectal motor function and offers the potential for characterizing abnormal mechanical properties of the rectal wall in disease.

We developed a cryomicrotome/imaging system that provides high resolution, high sensitivity block-face images of whole mice or excised organs, and applied it to a variety of biological applications. With this cryo-imaging system, we sectioned cryo-preserved tissues at 2-40 μm thickness and acquired high resolution brightfield and fluorescence images with microscopic in-plane resolution (as good as 1.2 μm). Brightfield images of normal and pathological anatomy show exquisite detail, especially in the abdominal cavity. Multi-planar reformatting and 3D renderings allow one to interrogate 3D structures. In this report, we present brightfield images of mouse anatomy, as well as 3D renderings of organs. For BPK mice model of polycystic kidney disease, we compared brightfield cryo-images and kidney volumes to MRI. The color images provided greater contrast and resolution of cysts as compared to in vivo MRI. We note that color cryo-images are closer to what a researcher sees in dissection, making it easier for them to interpret image data. The combination of field of view, depth of field, ultra high resolution and color/fluorescence contrast enables cryo-image volumes to provide details that cannot be found through in vivo imaging or other ex vivo optical imaging approaches. We believe that this novel imaging system will have applications that include identification of mouse phenotypes, characterization of diseases like blood vessel disease, kidney disease, and cancer, assessment of drug and gene therapy delivery and efficacy and validation of other imaging modalities.

Purpose: To evaluate photoacoustic CT spectroscopy (PCT-S) and dynamic contrast-enhanced CT (DCE-CT) ability to measure parameters - oxygen saturation and vascular physiology - associated with the intra-tumor oxygenation status. Material and Methods: Breast (VEGF165 enhance MCF-7) and ovarian (SKOV3x) cancer cells were implanted into the fat pads and flanks of immune deficient mice and allowed to grow to a diameter of 8-15 mm. CT was used to determine physiological parameters by acquiring a sequence of scans over a 10 minute period after an i.v. injection of a radio-opaque contrast agent (Isovue). These time-dependent contrast-enhanced curves were fit to a two-compartmental model determining tumor perfusion, fractional plasma volume, permeability-surface area produce, and fractional interstitial volume on a voxel-by-voxel basis. After which, the tumors were imaged using photoacoustic CT (Optosonics, Inc., Indianapolis, IN 46202). The near infrared spectra (700-910 nm) within the vasculature was fit to linear combination of measured oxy- and deoxy-hemoglobin blood samples to obtain oxygen saturation levels (SaO₂). Results: The PCT-S scanner was first calibrated using different samples of oxygenated blood, from which a statistical error ranging from 2.5-6.5% was measured and a plot of the hemoglobin dissociation curve was consistent with empirical formula. In vivo determination of tumor vasculature SaO₂ levels were measurably tracked, and spatially correlated to the periphery of the tumor. Tumor depend variations in SaO₂ - 0.32 (ovarian) and 0.60 (breast) - and in vascular physiology - perfusion, 1.03 and 0.063 mL/min/mL, and fractional plasma volume, 0.20 and 0.07 - were observed. Conclusion: Combined, PCT-S and CED-CT has the potential to measure intra-tumor levels of tumor oxygen saturation and vascular physiology, key parameters associated with hypoxia.

Optical tomography has been proposed as a promising technique for probing deep in tissue with many medical applications. Recently, the adaptation of fluorescent probes by the radiologists, gave rise to a new imaging tool in the area of molecular imaging. Optical tomography can, provide three-dimensional images of fluorescent concentrations inside living systems of sizes in the order of many cm. Our optical tomographer was based on a technique which is called Fluorescence Molecular Tomography (FMT) and can quantify fluorescent signals in mice. The imaging procedure is performed in a non-contact geometry so that living subjects of arbitrary shapes can be imaged with no fibers attached to them. We have developed a way to reconstruct the 3D surface of the subject and we use theoretical models to account for the propagation of the emerging signal in the free space. The system consists of a rotating sample holder and a CCD camera in combination with a laser-scanning device. An Argon-ion laser is used as the source and different filters are used for the detection of various fluorophores or fluorescing proteins. So far, we have observed of the distribution of GFP expressing T-lymphocytes in-vivo for the study of the function of the immune system in a murine model. Then we investigated the performance of the FMT setup to quantify the different amounts of migrated cells in the different organs by comparing our results with the FACS measurements. Further experiments included the measurement of the variations of the T cell's concentration in-vivo, over time.

Tissue microarray (TMA) technology allows rapid visualization of molecular targets in thousands of tissue specimens at a time and provides valuable information on expression of proteins within tissues at a cellular and sub-cellular level. TMA technology overcomes the bottleneck of traditional tissue analysis and allows it to catch up with the rapid advances in lead discovery. Studies using TMA on immunohistochemistry (IHC) can produce a large amount of images for interpretation within a very short time. Manual interpretation does not allow accurate quantitative analysis of staining to be undertaken. Automatic image capture and analysis has been shown to be superior to manual interpretation. The aims of this work is to develop a truly high-throughput and fully automated image capture and analysis system. We develop a robust colour segmentation algorithm using hue-saturation-intensity (HSI) colour space to provide quantification of signal intensity and partitioning of staining on high-throughput TMA. Initial segmentation results and quantification data have been achieved on 16,000 TMA colour images over 23 different tissue types.

The advent of optical molecular probes has taken optical imaging beyond approaches limited to intrinsic optical contrast mechanisms. Fluorophores are typically used as the source of contrast for optical molecular probes and the field of optical molecular imaging is concerned with measuring and quantifying their in vivo biodistribution and pharmacokinetics. Most optical molecular imaging systems are based on Continuous Wave (CW) approaches which enable rapid, full-body imaging of small animals and readily yield images of probe location, however quantification of probe concentration is challenging. Time Domain (TD) approaches, although more expensive and complicated than CW, provide more information to assist in determining the probe location and concentration. Moreover, the TD approach permits access to measuring the fluorophore lifetime which can be indicative of the probe's environment. The eXplore Optix^TM system, developed by ART (Canada) and distributed by GE Healthcare, has enabled TD optical molecular imaging of small animals in vivo and preliminary studies conducted with the system will be presented. In addition, the initial research and development of a full-field TD optical molecular imaging system incorporating a high-power laser for area illumination and a gated-intensified CCD camera for area detection will be presented.

Imaging of dynamic changes in blood parameters, functional brain imaging, and tumor imaging are the most advanced application areas of diffuse optical tomography (DOT). When dealing with the image reconstruction problem one is faced with the fact that near-infrared photons, unlike X-rays, are highly scattered when they traverse biological tissue. Image reconstruction schemes are required that model the light propagation inside biological tissue and predict measurements on the tissue surface. By iteratively changing the tissue-parameters until the predictions agree with the real measurements, a spatial distribution of optical properties inside the tissue is found. The optical properties can be related to the tissue oxygenation, inflammation, or to the fluorophore concentration of a biochemical marker. If the model of light propagation is inaccurate, the reconstruction process will lead to an inaccurate result as well. Here, we focus on difficulties that are encountered when DOT is employed for functional imaging of small tissue volumes, for example, in cancer studies involving small animals, or human finger joints for early diagnosis of rheumatoid arthritis. Most of the currently employed image reconstruction methods rely on the diffusion theory that is an approximation to the equation of radiative transfer. But, in the cases of small tissue volumes and tissues that contain low scattering regions diffusion theory has been shown to be of limited applicability Therefore, we employ a light propagation model that is based on the equation of radiative transfer, which promises to overcome the limitations.

Laminar Optical Tomography (LOT) is a new medical imaging modality for high-resolution, depth-resolved, functional imaging of superficial tissue such as rodent cortex, skin and the retina. LOT uses visible laser light to image to depths of >2mm (far deeper than microscopy) and is highly sensitive to absorption and fluorescence contrast, enabling spectroscopic functional information such as hemoglobin oxygenation to be imaged with 100-200 micron resolution. LOT has been used to image the hemodynamic response to stimulus in the somatosensory cortex of rats. The resulting three-dimensional (3D) images through the depth of the cortex can be used to delineate the arterial, capillary and venous responses, revealing new information about the intricacies of the oxygenation and blood flow dynamics related to neuronal activation. Additional applications of LOT are being explored, including the integration of 3D Voltage Sensitive Dye fluorescence imaging. LOT imaging uses a system similar to a confocal microscope, quickly scanning a focused beam of light over the surface of the tissue (~8Hz frame rate). Light is detected from both the focus of the scanning beam, and also at increasing distances from the beam's focus. This scattered light has penetrated more deeply into the tissue, and allows features at different depths to be distinguished. An algorithm that includes photon migration modeling of light scattering converts the raw data into 3D images. The motivation for functional optical imaging will be outlined, the basic principles of LOT imaging will be described, and the latest in-vivo results will be presented.

Simulated data is an important tool for evaluation of reconstruction and image processing algorithms in the frequent absence of ground truth, in-vivo data from living subjects. This is especially true in the case of dynamic PET studies, in which counting statistics of the volume can vary widely over the time-course of the acquisition. Realistic simulated data-sets which model anatomy and physiology, and make explicit the spatial and temporal image acquisition characteristics, facilitate experimentation with a wide range of the conditions anticipated in practice, and which can severely challenge algorithm performance and reliability. As a first example, we have developed a realistic dynamic FDG-PET data-set using the PET-SORTEO Monte Carlo simulation code and the MNI digital brain phantom. The phantom is a three-dimensional data-set that defines the spatial distribution of different tissues. Time activity curves were calculated using an impulse response function specified by generally accepted rate constants, convolved with an input function obtained by blood sampling, and assigned to grey and white matter tissue regions. We created a dynamic PET study using PET-SORTEO configured to simulate an ECAT Exact HR+. The resulting sinograms were reconstructed with all corrections, using variations of FBP and OSEM. Having constructed the dynamic PET data-sets, we used them to evaluate the performance of intensity-based registration as part of a tool for quantifying hyper/hypo perfusion with particular application to analysis of brain dementia scans, and a study of the stability of kinetic parameter estimation.

Nowadays best Brain Computer Interface (BCI) methods are based on invasive recording of electrical brain activity. Surface electrodes methods are not as accurate. This is partially due to the filtering of the signal by the skull and to the distance to the sources. Surprisingly methods for solving the EEG inverse problem have seldom been used to overcome these limitations. Inverse solution methods can be adapted to either pre-process the data or as a classification method. In this paper we study the application of well-known Inverse Solution methods to the BCI. Methods are the Minimum Norm method, two methods based on respectively a laplacian and a location prior, as well as two parametric methods based on subspace correlation. Data are processed with an inverse solution method. Then the data are classified by measuring the activation in preselected areas. Processing with Inverse method can improve the classification obtained outside of the skull by more than 10%. Furthermore these methods can be used without increasing the computation time. With our simple paradigm we obtained 85% of good classification.

The aim of this paper is to analyze spatiotemporal patterns of Event-related potential (ERP) in emotional processing by using fuzzy k-means clustering method to segment ERP data into microstates.108 pictures (categorized as positive, negative and neutral) were presented to 24 healthy, right-handed subjects while 128-channel EEG data were recorded. For each subject, 3 artifact-free ERPs were computed under each condition. A modified fuzzy k-mean clustering method based on shape similarity is applied to the grand mean ERPs and the statistical analysis is performed to define the significance of each segmentation map. In the results, positive and negative conditions showed different spatiotemporal patterns of ERP. The results were in accord with other emotional study by fMRI or PET.

The purpose of this paper is to study spatiotemporal patterns of neuronal activity in emotional processing by analysis of ERP data. 108 pictures (categorized as positive, negative and neutral) were presented to 24 healthy, right-handed subjects while 128-channel EEG data were recorded. An analysis of two steps was applied to the ERP data. First, principal component analysis was performed to obtain significant ERP components. Then LORETA was applied to each component to localize their brain sources. The first six principal components were extracted, each of which showed different spatiotemporal patterns of neuronal activity. The results agree with other emotional study by fMRI or PET. The combination of PCA and LORETA can be used to analyze spatiotemporal patterns of ERP data in emotional processing.

Diffusion tensor imaging has shown potential in providing information about the location of white matter tracts within the human brain. Based on this data, a novel approach is presented establishing connectivity between functional regions using pathfinding. The probability distribution function of the local tensor thereby controls the state space search performed by pathfinding. Additionally, it serves as an indicator for the reliability of the computed paths visualized by color encoding. Besides the capability to handle noisy data, the probabilistic nature of the approach is also able to cope with crossing or branching fibers. The algorithm thus guarantees to establish a connection between cortical regions and on the same hand provides information about the probability of the obtained connection. This approach is especially useful for investigating the connectivity between certain centers of the brain as demonstrated by reconstructed connections between motor and sensory speech areas.

We explore the use of scalar and multivariate autoregressive models to parameterize motion artifacts in fMRI time series. To do so, we acquire real fMRI data sets, measure rigid body motion in these data sets, and classify the type of observed motion in several categories such as random motion or motion correlated with activation. The measured motion sequences are then modeled and used to generate realistic image phantoms that can be used to validate fMRI data analysis packages. We compare phantoms generated with the original motion sequences and phantoms generated with simulated sequences. We show that both scalar and multivariate autoregressive models can be used to generate realistic motion sequences. An important difference between the two is the fact that multivariate models can capture correlations between motion parameters, which cannot be done with scalar models.

T₂ relaxation decay curves from in vivo human brain tissue are rarely mono-exponential due to both physiology and partial volume averaging. We propose a tri-exponential model, parametric fitting of the T₂ relaxation curve, restricting the range for the T₂ in each compartment, and estimating the probability of the existence of each of the components on a voxel-by-voxel basis. The model quantifies the T₂ into three discrete compartments: Myelin (T₂ short = 20-50 ms), White / Gray Matter (T₂ middle = 50-120 ms), and CSF (T₂ long = 120-500 ms). A constrained nonlinear minimization technique using subspace trust-region methods was implemented. A voxel-by-voxel analysis was performed, and for any given voxel, the three T₂ components were forced to lie within each compartment. However, the magnitude for each of these components was allowed to take any non-negative value including zero. As a result, if any component were absent, its magnitude would be zero and hence not contribute to the fit. Results from the processing of six healthy normal adults, imaged on a 3T magnet with clinically viable imaging protocols, have been presented and are shown to be in excellent agreement with reported values. This technique is robust and accurate and may potentially be useful in aiding clinical diagnosis and follow-up of patients with white matter abnormalities.

This study examined regional gray matter abnormalities across the whole brain in 19 patients with schizophrenia (12 males and 7 females), comparing with 11 normal volunteers (7 males and 4 females). The customized brain templates were created in order to improve spatial normalization and segmentation. Then automated preprocessing of magnetic resonance imaging (MRI) data was conducted using optimized voxel-based morphometry (VBM). The statistical voxel based analysis was implemented in terms of two-sample t-test model. Compared with normal controls, regional gray matter concentration in patients with schizophrenia was significantly reduced in the bilateral superior temporal gyrus, bilateral middle frontal and inferior frontal gyrus, right insula, precentral and parahippocampal areas, left thalamus and hypothalamus as well as, however, significant increases in gray matter concentration were not observed across the whole brain in the patients. This study confirms and extends some earlier findings on gray matter abnormalities in schizophrenic patients. Previous behavior and fMRI researches on schizophrenia have suggested that cognitive capacity decreased and self-conscious weakened in schizophrenic patients. These regional gray matter abnormalities determined through structural MRI with optimized VBM may be potential anatomic underpinnings of schizophrenia.

This work outlines a procedure for identifying brain structures not affected by prevailing global brain deformations. Such information may provide useful constraints for morphometric analysis methods used to measure neurodegenerative disease. The method proposed combines a global rigid registration and a localized block matching approach to analyze serial deformed images. Deformations from neurodegenerative morphological changes in tissue were simulated using a two dimensional finite element model based on porous-medial physics. Success of the registration/block-matching algorithm was determined if undeformed regions could be identified uniquely. Results indicated an ability to identify undeformed areas in five generated test cases. However, there were significant false-positive areas within the results which need further research to overcome.

We explored multiple image processing approaches by which to display the segmented adult brachial plexus in a three-dimensional manner. Magnetic resonance neurography (MRN) 1.5-Tesla scans with STIR sequences, which preferentially highlight nerves, were performed in adult volunteers to generate high-resolution raw images. Using multiple software programs, the raw MRN images were then manipulated so as to achieve segmentation of plexus neurovascular structures, which were incorporated into three different visualization schemes: rotating upper thoracic girdle skeletal frames, dynamic fly-throughs parallel to the clavicle, and thin slab volume-rendered composite projections.

Recent advances in imaging technologies, such as Magnetic Resonance Imaging (MRI), Positron Emission Tomography (PET) and Diffusion Tensor Imaging (DTI) have accelerated brain research in many aspects. In order to better understand the synergy of the many processes involved in normal brain function, integrated modeling and analysis of MRI, PET, and DTI is highly desirable. Unfortunately, the current state-of-art computational tools fall short in offering a comprehensive computational framework that is accurate and mathematically rigorous. In this paper we present a framework which is based on conformal parameterization of a brain from high-resolution structural MRI data to a canonical spherical domain. This model allows natural integration of information from co-registered PET as well as DTI data and lays the foundation for a quantitative analysis of the relationship between diverse data sets. Consequently, the system can be designed to provide a software environment able to facilitate statistical detection of abnormal functional brain patterns in patients with a large number of neurological disorders.

Uterine leiomyomas are the most common pelvic tumors in females. The efficacy of medical treatment is gauged by shrinkage of the size of these tumors. In this paper, we present a method to robustly segment the fibroids on MRI and accurately measure the 3D volume. Our method is based on a combination of fast marching level set and Laplacian level set. With a seed point placed inside the fibroid region, a fast marching level set is first employed to obtain a rough segmentation, followed by a Laplacian level set to refine the segmentation. We devised a scheme to automatically determine the parameters for the level set function and the sigmoid function based on pixel statistics around the seed point. The segmentation is conducted on three concurrent views (axial, coronal and sagittal), and a combined volume measurement is computed to obtain a more reliable measurement. We carried out extensive tests on 13 patients, 25 MRI studies and 133 fibroids. The segmentation result was validated against manual segmentation defined by experts. The average segmentation sensitivity (true positive fraction) among all fibroids was 84.6%, and the average segmentation specificity (1-false positive fraction) was 84.3%.

Previous studies have demonstrated the plausibility of using volumetric computerized tomography to provide an accurate representation and measurement of volume for pelvic osteolytic lesions following total hip joint replacement. These studies have been performed manually (or computed-assisted) by expert radiologists with the disadvantage of poor reproducibility of the experiment. The purpose of this work is to minimize the effect of user interaction in these experiments by introducing Laplacian level set methods in the volume segmentation process and using temporal articulated registration in order to follow the evolution of a lesion over time. Laplacian level set methods reduce the inter and intra-observer variability by attaching the segmented contour to edges defined in the image while keeping smoothness. The registration process allows the information of the lesion from the first visit to be used in the segmentation process of the current visit. This work compares the automated results on 7 volunteers versus the volume measured manually. Results have shown that the proposed technique is able to track osteolytic lesions and detect changes in volume over time. Intra-reader and inter-observer variabilities were reduced.

An automatic, simple, and image intensity standardization-based strategy for correcting background inhomogeneity in MR images is presented in this paper. Image intensities are first transformed to a standard intensity gray scale by a standardization process. Different tissue sample regions are then obtained from the standardized image by simply thresholding based on fixed intensity intervals. For each tissue region, a polynomial is fitted to the estimated discrete background intensity variation. Finally, a combined polynomial is determined and used for correcting the intensity inhomogeneity in the whole image. The above procedure is repeated on the corrected image iteratively until the size of the extracted tissue regions does not change significantly in two successive iterations. Intensity scale standardization is effected to make sure that the corrected image is not biased by the fitting strategy. The method has been tested on a number of simulated and clinical MR images. These tests and a comparison with the method of non-parametric non-uniform intensity normalization (N3) indicate that the method is effective for background intensity inhomogeneity correction and may have a slight edge over the N3 method.

Motion estimation is an essential processing step common to all Magnetic Resonance Elastography (MRE) methods. For dynamic techniques, the motion is obtained from a sinusoidal fit of the image phase at multiple, uniformly spaced relative phase offsets, φ, between the motion and the motion encoding gradients (MEGs). Generally, 4 to 8 uniformly spaced values of φ are used. We introduce a method, termed RME (reduced motion encodes), of reducing the number of relative phases required, thereby reducing the imaging time for an MRE acquisition. A frequency-domain algorithm was implemented using the Discrete Fourier Transform (DFT) to derive the general least-squares solution for the motion amplitude and phase given an arbitrary number of phase offsets. Simulation result shows that the noise level decreases as the number of φ increases. The decrease is largest when smaller numbers of φ are used and becomes less significant as the number increases. The minimum noise is obtained for a specific number, n, of φ when the phase is evenly distributed with interval π/n. Phantom studies show a similar trend with noise level. The resulting displacement images from different numbers of phase offsets are compared.

This paper explores the use of statistical parametric mapping, which has been widely used in brain imaging applications, in the assessment of morphological changes in rat articular cartilage during OA. This approach can not only detect but also localize changes, effectively zooming in statistically more informative areas and therefore maximizing the sensitivity. The analysis of in-vivo MR images of rat articular cartilage demonstrates that the new approach is at least as sensitive as the cartilage volume analysis.

Bicipital root and proximal tendon disorders are an important symptom generator in the shoulder. The accuracy of the diagnosis of many shoulder disorders visually without quantitative shape analysis is limited, motivating a clinical need for some ancillary method to access the proximal biceps. Because of the known inter-relationship of the bicipital groove (BG) with several types of disorders, we propose an approach to the 3D shape description of the BG that captures information relevant to disorders of the shoulder (e.g. width, depth, angles of walls, presence of spurs). Our approach is medial-axis based and captures intuitive aspects of shape such as thickness, bending, and elongation. Our proposed method overcomes the well-known problem of boundary sensitivity in the medial representation as it is applied to representation and analysis of BG shape. We give preliminary quantitative results indicating that this representation does capture shape variation within our experimental data, providing motivation to explore more sophisticated statistical analysis based on this representation in future work. We also provide a method for semi-automatic segmentation of the BG from computed tomography (CT) scans of the shoulder; an important precursor step to BG shape analysis.

Acquisition and analysis of three-dimensional movement of knee joint is desired in orthopedic surgery. We have developed two methods to obtain dynamic volume images of knee joint. One is a 2D/3D registration method combining a bi-plane dynamic X-ray fluoroscopy and a static three-dimensional CT, the other is a method using so-called 4D-CT that uses a cone-beam and a wide 2D detector. In this paper, we present two analyses of knee joint movement obtained by these methods: (1) transition of the nearest points between femur and tibia (2) principal component analysis (PCA) of six parameters representing the three dimensional movement of knee. As a preprocessing for the analysis, at first the femur and tibia regions are extracted from volume data at each time frame and then the registration of the tibia between different frames by an affine transformation consisting of rotation and translation are performed. The same transformation is applied femur as well. Using those image data, the movement of femur relative to tibia can be analyzed. Six movement parameters of femur consisting of three translation parameters and three rotation parameters are obtained from those images. In the analysis (1), axis of each bone is first found and then the flexion angle of the knee joint is calculated. For each flexion angle, the minimum distance between femur and tibia and the location giving the minimum distance are found in both lateral condyle and medial condyle. As a result, it was observed that the movement of lateral condyle is larger than medial condyle. In the analysis (2), it was found that the movement of the knee can be represented by the first three principal components with precision of 99.58% and those three components seem to strongly relate to three major movements of femur in the knee bend known in orthopedic surgery.

Recent advances in medical research hypothesize that certain body fat, in addition to having a classical role of energy storage, may also have mechanical function. In particular, we analyzed the infrapatellar fat pad of Hoffa using 3D CT images of the knee at multiple angles to determine how the fat pad changes shape as the knee bends and whether the fat pad provides cushioning in the knee joint. The images were initially processed using a median filter then segmented using a region growing technique to isolate the fat pad from the rest of the knee. Next, rigid registration was performed to align the series of images to match the reference image. Finally, multi-resolution FEM registration was completed between the aligned images. The resulting displacements fields were used to determine the local volume change of the fat pad as the knee bends from extension to flexion through different angles. This multi-angle analysis provides a finer description of the intermediate deformations compared to earlier work, where only a pair of images (full extension and flexion) was analyzed.

This research proposes the deformation model of organs for the development of the medical training system using Virtual Reality (VR) technology. First, the proposed model calculates the strains of coordinate axis. Secondly, the deformation is obtained by mapping the coordinate of the object to the strained coordinate. We assume the beams in the coordinate space to calculate the strain of the coordinate axis. The forces acting on the object are converted to the forces applied to the beams. The bend and the twist of the beams are calculated based on the theory of structural mechanics. The bend is derived by the finite element method. We propose two deformation methods which differ in the position of the beams in the coordinate space. One method locates the beams along the three orthogonal axes (x, y, z). Another method locates the beam in the area where the deformation is large. In addition, the strain of the coordinate axis is attenuated in proportion to the distance from the point of action to consider the attenuation of the stress which is a viscoelastic feature of the organs. The proposed model needs less computational cost compared to the conventional deformation method since our model does not need to divide the object into the elasticity element. The proposed model was implemented in the laparoscopic surgery training system, and a real-time deformation can be realized.

This paper is about the quantitative prediction of the long term outcome of the endovascular coiling treatment of a patient's cerebral aneurysm. It is generally believed that the local hemodynamic properties of the patient's cerebral arteries are strongly influencing the origin and growth of aneurysms. We describe our approach: modelling the flow in a 3D Rotational Angiography (3DRA) reconstruction of the aneurysms including supplying and draining blood vessels, in combination with simulations and experiments of artificial blood vessel phantom constructs and measurements. The goal is to obtain insight in the observed phenomena to support the diagnostic decision process in order to predict the outcome of the intervention with possible simulation of the flow alternation due to the pertinent intervention.

Recent advances in medical imaging, computational methods, and biomechanics hold great promise for engineering-based decision making in clinical practice. Towards patient-specific modeling, however, we need to synthesize better the separate advances in computational biofluid mechanics and arterial wall mechanics. In this paper, we propose a mathematical model of growing fusiform aneurysms that is able to test multiple competing hypotheses with regard to the production, removal, and organization of intramural collagen, and thus to predict their consequences in enlargement and changes in material properties of the lesion. To apply this model to realistic cases, including fluid-solid interactions, we also need to develop a method to exploit current advances in computational biofluid mechanics. Thus, we describe a method to represent highly nonlinear and anisotropic material behaviors within a linearized constitutive equation commonly employed in fluid-structure simulations of blood flow in deformable arteries.

In recent years, simulations of the blood flow and the wall mechanics in the vascular system with patient-specific boundary conditions by using computational fluid dynamics (CFD) and computational solid mechanics (CSM) have gained significant interest. A common goal of such simulations is to help predict the development of vascular diseases over time. However, the validity of such simulations and therefore the validity of the predictions are often questioned by physicians. The aim of the research reported in this paper is to validate CFD simulations performed on patient-specific models of abdominal aorta aneurysms (AAAs) using patient-specific blood velocity inflow profiles. Patient-specific AAA geometries were derived from images originating from Computed Tomography (CT) or Magnetic Resonance (MR) imaging. Patient-specific flow profiles were measured with Phase-Contrast MR imaging (Quantitative flow, Qflow). Such profiles, determined at the inflow site of the AAA, were used as inflow boundary condition for CFD simulations. Qflow images that were taken on a number of planes along the AAA were used for the validation of the simulation results. To compare the measured with the simulated flow we have generated synthetic Qflow images from the simulated velocities on cut-planes positioned and oriented according to the planes of the validation images. The comparison of the real with the simulated flow profiles was performed visually and by quantitatively comparing flow values on cross sections of the AAA in the measured and the synthetic Qflow images. In a preliminary study on two patients we found a reasonable agreement between the measured and the simulated flow profiles.

Finite element wall stress simulations on patient-specific models of abdominal aortic aneurysm (AAA) may provide a better rupture risk predictor than the currently used maximum transverse diameter. Calcifications in the wall of AAA lead to a higher maximum wall stress and thus may lead to an elevated rupture risk. The reported material properties for calcifications and the material properties actually used for simulations show great variation. Previous studies have focused on simplified modelling of the calcification shapes within a realistic aneurysm shape. In this study we use an accurate representation of the calcification geometry and a simplified model for the AAA. The objective of this approach is to investigate the influence of the calcification geometry, the material properties and the modelling approach for the computed peak wall stress. For four realistic calcification shapes from standard clinical CT images of AAA, we performed simulations with three distinct modelling approaches, at five distinct elasticity settings. The results show how peak wall stress is sensitive to the material properties of the calcifications. For relatively elastic calcifications, the results from the different modelling approaches agree. Also, for relatively elastic calcifications the computed wall stress in the tissue surrounding the calcifications shows to be insensitive to the exact calcification geometry. For stiffer calcifications the different modelling approaches and the different geometries lead to significantly different results. We conclude that an important challenge for future research is accurately estimating the material properties and the rupture potential of the AAA wall including calcifications.

Stenting may provide a new, less invasive therapeutic option for cerebral aneurysms. However, a conventional porous stent may be insufficient in modifying the blood flow for clinical aneurysms. We designed an asymmetric stent consisting of a low porosity patch welded onto a porous stent for an anterior cerebral artery aneurysm of a specific patient geometry to block the strong inflow jet. To evaluate the effect of the patch on aneurysmal flow dynamics, we "virtually" implanted it into the patient's aneurysm geometry and performed Computational Fluid Dynamics (CFD) analysis. The patch was computationally deformed to fit into the vessel lumen segmented from the patient CT reconstructions. After the flow calculations, a patch with the same design was fabricated using laser cutting techniques and welded onto a commercial porous stent, creating a patient-specific asymmetric stent. This stent was implanted into a phantom, which was imaged with X-ray angiography. The hemodynamics of untreated and stented aneurysms were compared both computationally and experimentally. It was found from CFD of the patient aneurysm that the asymmetric stent effectively blocked the strong inflow jet into the aneurysm and eliminated the flow impingement on the aneurysm wall at the dome. The impact zone with elevated wall shear stress was eliminated, the aneurysmal flow activity was substantially reduced, and the flow was considerably reduced. Experimental observations corresponded well qualitatively with the CFD results. The demonstrated asymmetric stent could lead to a new minimally invasive image guided intervention to reduce aneurysm growth and rupture.

Diameters (or areas) of vessel cross-sections provide useful information for diagnosis and surgery planning. However, the ordinary centerline-perpendicular cross-sections are often inappropriate to use because the centerline may include unwanted local curvatures in irregular or asymmetric regions and high curvatures in sharply bended regions. In this paper, we try to improve the accuracy of vessel cross-section measurement by properly adjusting the centerline. To alleviate local curvatures in the centerline while preserving the global shape faithfully, we register a deformable cylindrical model onto the vessel lumen, and subsequently adopt the axis of the registered model as the adjusted centerline for determining cross-sections. In addition, by introducing the electric field model, we prevent undesirable intersection of cross-sections that is often found in sharply bended regions. Experiments are performed using various synthesized images that simulate abnormal vessels with stenoses or aneurysms. The results show that the registration process successfully eliminates unwanted local curvatures while preserving the global shape of the vessel, and obtained cross-sections do not intersect each other even in the region of high curvature.

X-ray coronary angiography is widely used to determine the presence of a stenosis. This paper discusses an approach towards the detection of the functional severity of a stenosis using the relative velocity of the contrast agent. The velocity of the contrast is measured using the arrival time at several locations on a coronary artery. This is done by placing multiple Regions Of Interest(ROI) equally spaced on the artery. The location of these ROIs varies in time because of the cardiac motion. Therefore, an artery tracing and tracking algorithm is used to estimate the location of the ROIs in time. The arrival time of the contrast can be estimated by measuring the image intensity in these ROIs. Using the arrival times in several ROIs, a qualitative velocity can be estimated. Altering the velocity of the blood pharmacologically, by inducing hyperemic conditions, results in a qualitative change in velocity detected by the algorithm. No change in velocity may indicate a severe flow limiting stenosis.

Via angiographic injection, blood vessels become visible in X-ray imaging. Based on dynamic bolus tracking and static vessel volume depiction, this paper presents a three-dimensional (3D) imaging method for blood flow measurement. After angiographic injection, the bolus motion in vessels in a scan field of view can be described by wash-in, equilibrium, and wash-out phases in the order of time. The cone-beam scanning produces a sequence of projection images, of which the wash-in images are used for bolus tracking, and the equilibrium images for vessel volume reconstruction. From a vessel volume, not only can we depict the vessel diameter by digital anatomical analysis, but also extract vessel centerlines by volumetric vessel segmentation and 3D skeletonization. We assume that bolus travels along the vessel centerlines, and the 3D bolus passageways can be time ticked by consulting the bolus motion in a sequence of wash-in projection images. By splitting this sequence into two subsequences (different by a delay of a few frames) and considering them as two sequences of two-view image pairs, we can calculate a 3D bolus passageway by two-view stereo reconstruction and then correct the excursions by a nudge algorithm (3D bolus point adjustment in reference to 3D vessel skeleton). Alternatively, by cone-beam reprojecting the 3D skeleton and consulting the 2D bolus motion manifested in wash-in projection images, we can tick off the 3D skeleton by time and thereby calculating bolus pathlength and flow velocity. Numerical simulations and phantom experiments are reported.

Cardiovascular disease is considered the leading cause of death in the US, accounting for 38% of all deaths. There are gender differences in the size of coronary arteries and in the character and location of atherosclerotic lesions that affect the detection of coronary artery disease with the medical imaging modalities currently used (e.g. angiography, computed tomography). These differences also affect the safety and effectiveness of image-guided interventions using therapeutic devices. For the optimization of the medical imaging modalities used for this specific task we require the generation of clinically-realistic, gender-specific images of healthy and pathological coronary angiograms. For this purpose we have created a gender-specific statistical model of a pathological coronary artery tree. Starting from "healthy" heart-phantoms created from high resolution CT scans of cadaver hearts of both genders, the model uses prevalence data obtained from clinical studies of patients with significant (>50% stenosis) coronary artery disease (CAD). The model determines the plaque deposit locations and character (length, percent stenosis) for each case, based on a flow model. These data are then used to generate artificially diseased artery trees, embedded in a gender-specific torso model. Using an x-ray and optical photon Monte-Carlo simulation program, we then generate simulated angiograms exhibiting realistic disease patterns. The severity of each angiogram is determined from a set of rules that combines the geometrically increasing severity of lesions, the cumulative effects of multiple obstructions, the significance of their locations, the modifying influence of the collaterals, and the size and quality of the distal vessels. The simulated angiograms will consequently be read by model and human observers. The probability of detection derived in combination with the severity score will be used as a figure of merit for the patient- and gender-specific optimization of the imaging modality under investigation.

Arterial stenoses may be obscured by calcified atherosclerotic plaque in computed tomographic angiography (CTA). A technique of subtraction computed tomographic angiography (sCTA) for calcification removal is proposed and evaluated in a preliminary manner. In the proposed sCTA method, the examination includes a pre-contrast computed tomogram (pcCT) and a CTA. The pcCT is performed using the CTA scan protocol. Subtraction of the registered pcCT from the CTA is performed after calcifications in the pcCT are registered to the CTA using a piecewise-rigid transformation model. The registration is based on a maximum-cancellation cost-function. Points at the boundary of the artery in the CTA are given a greater weight in the cost function than those towards the center of the artery. sCTA was evaluated using a calcified-artery phantom whose dimensions approximate those of the superficial femoral artery. The phantom represented both calcified plaque surrounding stenotic segments of the artery. pcCT and CTA's were obtained on a 4-multidetector-row CT system with 1.25-mm slice thickness and 0.7-mm in-plane resolution. The phantom was slightly displaced between the pcCT and the CTA. sCTA closely resembled a gold-standard image of the phantom that was obtained with the calcification material removed. The sCTA accurately demonstrated the degree of stenosis and artifacts in the sCTA were minimal. This study demonstrates in a preliminary manner that sCTA is feasible.

A vascular network is often represented by a Reeb graph, which is a topological skeleton, and graph theory has been widely applied to analyze properties of a vascular network. A Reeb graph model for a vascular network is obtained by assigning the branch points of the network to be the vertices of the graph and the vessels between branch points to be the edges of the graph. Vascular networks develop by way of angiogenesis, a growth process that involves the biological mechanisms of vessel sprouting (budding) and splitting (intussusception). Vascular networks develop by way of two biological mechanisms of vessel sprouting (budding) and splitting (intussusception). According to a graph theory modeling of two vascular network growth mechanisms, all nodes in the Reeb graph must be cubic in degree except for two special nodes: the afferent (A) and efferent (E) nodes. We define that a vascular network is cubic if all internal nodes are cubic in degree. We consider six normal adult rat renal glomerular networks and use their reeb graphs already constructed and published in the literature. We observe that five of them contain internal vertices of degree higher than three. Branch points in vascular networks may appear to be of a higher degree if the imaging resolution cannot differentiate between blood vessels that are very close in proximity. Here, we propose a random graph theory model that edits a non-cubic vascular network into a cubic graph. We observe that the edited cubic graph from a non-cubic vascular network has the similar size and order as the one cubic vascular network.

Initial animal study for quantifying myocardial physiology through contrast-enhanced dynamic x-ray CT suggested that beam hardening is one of the limiting factors for accurate regional physiology measurement. In this study, a series of simulations were performed to investigate its deterioration effects and two correction algorithms were adapted to evaluate for their efficiency in improving the measurements. The simulation tool consists of a module simulating data acquisition of a real polyenergetic scanner system and a heart phantom consisting of simple geometric objects representing ventricles and myocardium. Each phantom component was modeled with time-varying attenuation coefficients determined by ideal iodine contrast dynamic curves obtained from experimental data or simulation. A compartment model was used to generate the ideal myocardium contrast curve using physiological parameters consistent with measured values. Projection data of the phantom were simulated and reconstructed to produce a sequence of simulated CT images. Simulated contrast dynamic curves were fitted to the compartmental model and the resultant physiological parameters were compared with ideal values to estimate the errors induced by beam hardening artifacts. The simulations yielded similar deterioration patterns of contrast dynamic curves as observed in the initial study. Significant underestimation of left ventricle curves and corruption of regional myocardium curves result in systematic errors of regional perfusion up to approximately 24% and overestimates of fractional blood volume (f_iv) up to 13%. The correction algorithms lead to significant improvement with errors of perfusion reduced to 7% and errors of f_iv within 2% which shows promise for more robust myocardial physiology measurement.

Myocardial infarction results in myocardial necrosis, usually caused by an imbalance in the oxygen supply and demand to myocardial tissue. To our knowledge there is no technique that can provide quantitative direct information concerning the intensity, extent and location of the infarction. Contraction forces generated by cardiac tissues represent a quantitative and direct measure of the myocardial functionality, since it is expected that infarcted tissue generate little or no contraction force. Our objective is to develop a biomechanics based reconstruction technique to image myocardial contraction forces, for the purpose of assessing the viability of cardiac tissues. This technique is designed to reconstruct the contraction forces by inverting myocardial tissue displacement data acquired throughout heart beat cycles using conventional imaging techniques. Recognizing that myocardial contraction force distribution is 3D, we assumed an axisymmetric myocardial geometry to demonstrate the concept. With this assumption, the inversion algorithm was developed and implemented in 2D space. As a preliminary analysis, a simulation involving a 2D representation of myocardial wall tissue was carried out. The tissue was modeled as a homogeneous material with isotropic and linear elastic material properties. Assuming an axisymmetric contraction force distribution, a finite element analysis was performed on the tissue model, and a 2D displacement field was generated. The developed inversion algorithm was then employed to reconstruct the force distribution, which was ultimately compared to the original force field. The reconstruction error, estimated as the difference between the two force fields and normalized by the magnitude of the reference distribution, averaged to +/-10%. Results demonstrate that our myocardial contraction force reconstruction algorithm is reasonably accurate and robust.

Harmonic phase (HARP) MRI is used to measure myocardial motion and strain from tagged MR images. HARP MRI uses limited number of samples from the spectrum of the tagged images to reconstruct motion and strain. The HARP strain maps, however, suffer from artifacts that limit the accuracy of the computations and degrade the appearance of the strain maps. Causes of these, so called 'zebra', artifacts include image noise, Gibbs ringing, and interference from other Fourier spectral peaks. Computing derivatives of the HARP phase, which are needed to estimate strain, further accentuates these artifacts. Previous methods to reduce these artifacts include 1-D and 2-D nonlinear filtering of the HARP derivatives, and a 2-D linear filtering of unwrapped HARP phase. A common drawback among these methods is the lack of proper segmentation of the myocardium from the blood pool. Because of the lack of segmentation, the noisy phase values from the blood pool enter into the computation in the smoothed strain maps, which causes artifacts. In this work, we propose a smoothing method based on anisotropic diffusion that filters the HARP derivatives strictly within the myocardium without the need for prior segmentation. The information about tissue geometry and the strain distribution is used to restrict the smoothing to within the myocardium, thereby ensuring minimum distortion of the final strain map. Preliminary results demonstrate the ability of anisotropic diffusion for better artifact reduction and lesser strain distortion than the existing methods.

In this paper, an improved framework for estimation of 3-D left-ventricular deformations from tagged MRI is presented. Contiguous short- and long-axis tagged MR images are collected and are used within a 4-D B-Spline based deformable model to determine 4-D displacements and strains. An initial 4-D B-spline model fitted to sparse tag line data is first constructed by minimizing a 4-D Chamfer distance potential-based energy function for aligning isoparametric planes of the model with tag line locations; subsequently, dense virtual tag lines based on 2-D phase-based displacement estimates and the initial model are created. A final 4-D B-spline model with increased knots is fitted to the virtual tag lines. From the final model, we can extract accurate 3-D myocardial deformation fields and corresponding strain maps which are local measures of non-rigid deformation. Lagrangian strains in simulated data are derived which show improvement over our previous work. The method is also applied to 3-D tagged MRI data collected in a canine.

The purpose of the study was to decrease the variability of computed tomographic airway measurements. We to developed and evaluated a novel computer scheme to automatically segment airways depicted on chest CT examinations at the level of the lobar and segmental bronchi and to decrease. The computer scheme begins with manual selection of a seed point within the airway from which the airway wall and lumen are automatically segmented and airway pixels were assigned full or partial membership to the lumen or wall. Airway pixels not assigned full membership to the lumen (< -900 HU) or wall (> 0 HU) were assigned partial membership to the lumen and wall. In fifteen subjects with no visible signs of emphysema and a range of pulmonary obstruction from none to severe, airway measures were compared to pulmonary function parameters in a rank order analysis to evaluate measuring a single airway versus multiple airways. The quality of the automated airway segmentation was visually acceptable. The Pearson Correlation coefficients for the ranking of FEV₁ versus wall area percent (percent of total airway size) and FVC versus wall area percent were 0.164 and 0.175 for a single measurement, respectively, and were 0.243 and 0.239 for multiple measurements, respectively. Our preliminary results suggest that averaging the measurements from multiple airways may improve the relation between airway measures and lung function compared to measurement from a single airway, which improve quantification of airway remodeling in COPD patients.

The fraction of lung voxels below a pixel value "cut-off" has been correlated with pathologic estimates of emphysema. We performed a "standard" quantitative CT (QCT) lung analysis using a -950 HU cut-off to determine the volume fraction of emphysema (below the cut-off) and a "corrected" QCT analysis after removing small group (5 and 10 pixels) of connected pixels ("blobs") below the cut-off. CT examinations two dataset of 15 subjects each with a range of visible emphysema and pulmonary obstruction were acquired at "low-dose and conventional dose reconstructed using a high-spatial frequency kernel at 2.5 mm section thickness for the same subject. The "blob" size (i.e., connected-pixels) removed was inversely related to the computed fraction of emphysema. The slopes of emphysema fraction versus blob size were 0.013, 0.009, and 0.005 for subjects with both no emphysema and no pulmonary obstruction, moderate emphysema and pulmonary obstruction, and severe emphysema and severe pulmonary obstruction, respectively. The slopes of emphysema fraction versus blob size were 0.008 and 0.006 for low-dose and conventional CT examinations, respectively. The small blobs of pixels removed are most likely CT image artifacts and do not represent actual emphysema. The magnitude of the blob correction was appropriately associated with COPD severity. The blob correction appears to be applicable to QCT analysis in low-dose and conventional CT exams.

Recently, due to aging and smoking, emphysema patients are increasing. The restoration of alveolus which was destroyed by emphysema is not possible, thus early detection of emphysema is desired. We describe a quantitative algorithm for extracting emphysematous lesions and quantitatively evaluate their distribution patterns using low dose thoracic 3-D CT images. The algorithm identified lung anatomies, and extracted low attenuation area (LAA) as emphysematous lesion candidates. Applying the algorithm to 100 thoracic 3-D CT images and then by follow-up 3-D CT images, we demonstrate its potential effectiveness to assist radiologists and physicians to quantitatively evaluate the emphysematous lesions distribution and their evolution in time interval changes.

A segmentation method is a mandatory pre-processing step in many automated or semi-automated analysis tasks such as region identification and densitometric analysis, or even for 3D visualization purposes. In this work we present a fully automated volumetric pulmonary segmentation algorithm based on intensity discrimination and morphologic procedures. Our method first identifies the trachea as well as primary bronchi and then the pulmonary region is identified by applying a threshold and morphologic operations. When both lungs are in contact, additional procedures are performed to obtain two separated lung volumes. To evaluate the performance of the method, we compared contours extracted from 3D lung surfaces with reference contours, using several figures of merit. Results show that the worst case generally occurs at the middle sections of high resolution CT exams, due the presence of aerial and vascular structures. Nevertheless, the average error is inferior to the average error associated with radiologist inter-observer variability, which suggests that our method produces lung contours similar to those drawn by radiologists. The information created by our segmentation algorithm is used by an identification and representation method in pulmonary emphysema that also classifies emphysema according to its severity degree. Two clinically proved thresholds are applied which identify regions with severe emphysema, and with highly severe emphysema. Based on this thresholding strategy, an application for volumetric emphysema assessment was developed offering new display paradigms concerning the visualization of classification results. This framework is easily extendable to accommodate other classifiers namely those related with texture based segmentation as it is often the case with interstitial diseases.

Automatic extraction of the tracheobronchial tree from high resolution CT data serves visual inspection by virtual endoscopy as well as computer aided measurement of clinical parameters along the airways. The purpose of this study is to show the feasibility of automatic extraction (segmentation) of the airway tree even in ultra-low-dose CT data (5-10 mAs), and to compare the performance of the airway extraction between ultra-low-dose and standard-dose (70-100 mAs) CT data. A direct performance comparison (instead of a mere simulation) was possible since for each patient both an ultra-low-dose and a standard-dose CT scan were acquired within the same examination session. The data sets were recorded with a multi-slice CT scanner at the Charite university hospital Berlin with 1 mm slice thickness. An automated tree extraction algorithm was applied to both the ultra-low-dose and the standard-dose CT data. No dose-specific parameter-tuning or image pre-processing was used. For performance comparison, the total length of all visually verified centerlines of each tree was accumulated for all airways beyond the tracheal carina. Correlation of the extracted total airway length for ultra-low-dose versus standard-dose for each patient showed that on average in the ultra-low-dose images 84% of the length of the standard-dose images was retrieved.

Multi-slice CT technology was developed, so, we can get clear contrast images and thin slice images. But doctors need to diagnosis many image, thus their load increases. Therefore, development of the algorithm that analyses lung internal-organs is expected. When doctors diagnose lung internal-organs, they understand it. So, detailed analyze of lung internal-organs is applicant to early detection of a nodule. Especially, analyzing bronchus provides that useful information of detection of airway disease and classification of the pulmonary vein and artery. In this paper, we describe a method for automated anatomical labeling algorithm of bronchial branches based on Multi-Slice CT images.

Multi-slice helical CT technology has been developed. Unlike the conventional helical CT, we can obtain CT images of two or more slices in 1 time scan. Therefore, we can get many images with a clear contrast and thin slice images in one time of scanning. The purpose of this presentation is to evaluate the proposed automatic extraction bronchus and pulmonary vein and artery on multi-slice CT images. The bronchus is extracted by application with region growing technique and the morphological filters, 3D distance transformation. These results indicate that the proposed algorithm provides the ability to accurately develop an automatic extraction algorithm of the bronchus on multi-slice CT images. In this report, we used pulmonary vein and artery marked by the doctor, It aims to discover an amount of the feature necessary for classifying the pulmonary vein and artery by using the anatomical feature. The classification of the pulmonary vein and artery is thought to be a necessary information for tumor's benign or malignity judgment. In this report, the amount of the feature in which the flow of the automation is based is analyzed by using three dimension images of pulmonary vein and artery and bronchus obtained by the specialized physician's marking.

Recently-developed dynamic flat-panel detector (FPD) with a large field of view is possible to obtain breathing chest radiographs, which provide respiratory kinetics information. This study was performed to investigate the ability of dynamic chest radiography using FPD to quantify relative ventilation according to respiratory physiology. We also reported the results of primary clinical study and described the possibility of clinical use of our method. Dynamic chest radiographs of 12 subjects involving abnormal subjects during respiration were obtained using a modified FPD system (30 frames in 10 seconds). Imaging was performed in three different positions (standing, and right and left decubitus positions) to change the distribution of local ventilation by changing the lung's own gravity in each area. The distance from the lung apex to the diaphragm (abbr. DLD) was measured by the edge detection technique for use as an index of respiratory phase. We measured pixel values in each lung area and calculated correlation coefficients with DLD. Differences in the pixel values between the maximum inspiratory and expiratory frame were calculated, and the trend of distribution was evaluated by two-way analysis of variance. Pixel value in each lung area was strongly associated with respiratory phase and its time variation and distribution were consistent with known properties in respiratory physiology. Dynamic chest radiography using FPD combined with our computerized methods was capable of quantifying relative amount of ventilation during respiration, and of detecting regional differences in ventilation. In the subjects with emphysema, areas with decreased respiratory changes in pixel value are consisted with the areas with air trapping. This method is expected to be a useful novel diagnostic imaging method for supporting diagnosis and follow-up of pulmonary disease, which presents with abnormalities in local ventilation.

Idiopathic pulmonary fibrosis (IPF, also known as Idiopathic Usual Interstitial Pneumontis, pathologically) is a progressive diffuse lung disease which has a median survival rate of less than four years with a prevalence of 15-20/100,000 in the United States. Global function changes are measured by pulmonary function tests and the diagnosis and extent of pulmonary structural changes are typically assessed by acquiring two-dimensional high resolution CT (HRCT) images. The acquisition and analysis of volumetric high resolution Multi-Detector CT (MDCT) images with nearly isotropic pixels offers the potential to measure both lung function and structure. This paper presents a new approach to three dimensional lung image analysis and classification of normal and abnormal structures in lungs with IPF.

To create a repository of clinical data, CT images and tissue samples and to more clearly understand the pathogenetic features of pulmonary fibrosis and emphysema, the National Heart, Lung, and Blood Institute (NHLBI) launched a cooperative effort known as the Lung Tissue Resource Consortium (LTRC). The CT images for the LTRC effort must contain accurate CT numbers in order to characterize tissues, and must have high-spatial resolution to show fine anatomic structures. This study was performed to optimize the CT image reconstruction algorithms to achieve these criteria. Quantitative analyses of phantom and clinical images were conducted. The ACR CT accreditation phantom containing five regions of distinct CT attenuations (CT numbers of approximately -1000 HU, -80 HU, 0 HU, 130 HU and 900 HU), and a high-contrast spatial resolution test pattern, was scanned using CT systems from two manufacturers (General Electric (GE) Healthcare and Siemens Medical Solutions). Phantom images were reconstructed using all relevant reconstruction algorithms. Mean CT numbers and image noise (standard deviation) were measured and compared for the five materials. Clinical high-resolution chest CT images acquired on a GE CT system for a patient with diffuse lung disease were reconstructed using BONE and STANDARD algorithms and evaluated by a thoracic radiologist in terms of image quality and disease extent. The clinical BONE images were processed with a 3 x 3 x 3 median filter to simulate a thicker slice reconstructed in smoother algorithms, which have traditionally been proven to provide an accurate estimation of emphysema extent in the lungs. Using a threshold technique, the volume of emphysema (defined as the percentage of lung voxels having a CT number lower than -950 HU) was computed for the STANDARD, BONE, and BONE filtered. The CT numbers measured in the ACR CT Phantom images were accurate for all reconstruction kernels for both manufacturers. As expected, visual evaluation of the spatial resolution bar patterns demonstrated that the BONE (GE) and B46f (Siemens) showed higher spatial resolution compared to the STANDARD (GE) or B30f (Siemens) reconstruction algorithms typically used for routine body CT imaging. Only the sharper images were deemed clinically acceptable for the evaluation of diffuse lung disease (e.g. emphysema). Quantitative analyses of the extent of emphysema in patient data showed the percent volumes above the -950 HU threshold as 9.4% for the BONE reconstruction, 5.9% for the STANDARD reconstruction, and 4.7% for the BONE filtered images. Contrary to the practice of using standard resolution CT images for the quantitation of diffuse lung disease, these data demonstrate that a single sharp reconstruction (BONE/B46f) should be used for both the qualitative and quantitative evaluation of diffuse lung disease. The sharper reconstruction images, which are required for diagnostic interpretation, provide accurate CT numbers over the range of -1000 to +900 HU and preserve the fidelity of small structures in the reconstructed images. A filtered version of the sharper images can be accurately substituted for images reconstructed with smoother kernels for comparison to previously published results.

We propose a new subtraction technique for accurately imaging lung perfusion and efficiently detecting pulmonary embolism in chest MDCT angiography. Our method is composed of five stages. First, optimal segmentation technique is performed for extracting same volume of the lungs, major airway and vascular structures from pre- and post-contrast images with different lung density. Second, initial registration based on apex, hilar point and center of inertia (COI) of each unilateral lung is proposed to correct the gross translational mismatch. Third, initial alignment is refined by iterative surface registration. For fast and robust convergence of the distance measure to the optimal value, a 3D distance map is generated by the narrow-band distance propagation. Fourth, 3D nonlinear filter is applied to the lung parenchyma to compensate for residual spiral artifacts and artifacts caused by heart motion. Fifth, enhanced vessels are visualized by subtracting registered pre-contrast images from post-contrast images. To facilitate visualization of parenchyma enhancement, color-coded mapping and image fusion is used. Our method has been successfully applied to ten patients of pre- and post-contrast images in chest MDCT angiography. Experimental results show that the performance of our method is very promising compared with conventional methods with the aspects of its visual inspection, accuracy and processing time.

With perfusion magnetic resonance imaging (pMRI), perfusion describes the amount of blood passing through a block of tissue in a certain period of time. In pMRI, the tissue having more blood passing through will show higher intensity value as more contrast-labeled blood arrives. Perfusion reflects the delivery of essential nutrients to a block of tissue, and is an important parameter for the tissue status. Considering solitary pulmonary nodules (SPN), perfusion differences between malignant and benign nodules have been studied by different techniques. Much effort has been put into its characterization. In this paper, we proposed and implemented extraction of the SPN time intensity profile to measure blood delivery to solitary pulmonary nodules, describing their perfusion effects. In this method, a SPN time intensity profile is created based on intensity values of the solitary pulmonary nodule in lung pMRI images over time. This method has two steps: nodule tracking and profile clustering. Nodule tracking aligns the solitary pulmonary nodule in pMRI images taken at different time points, dealing with nodule movement resulted from breathing and body movement. Profile clustering implements segmentation of the nodule region and extraction of the time intensity profile of a solitary pulmonary nodule. SPN time intensity profiles reflect patterns of blood delivery to solitary pulmonary nodules, giving us a description of perfusion effect and indirect evidence of tumor angiogenesis. Analysis on SPN time intensity profiles will help the diagnosis of malignant nodules for early lung cancer detection.

In research and development of computer-aided differential diagnosis using thoracic CT images, there is now widespread interest in the use of nodule doubling time for measuring the volumetric changes of pulmonary nodule. The evolution pattern of each nodule might depend on the CT density distribution pattern inside nodule such as pure GGO, mixed GGO, or solid nodules. This paper presents a computerized approach to measure nodule density variation inside small pulmonary nodule using CT images. The approach consists of five steps: (1) nodule segmentation, (2) computation of CT density histogram, (3) nodule categorization (α, β, γ, δ, and ε) based on CT density histogram, (4) computation of doubling time based on CT density histogram, and (5) classification between benign and malignant cases. Using our dataset of follow-up scans of pulmonary nodules, we evaluated evaluation patterns of nodules on the basis of the predominant five nodule categorizations and designed the classification approach between benign and malignant cases. The preliminary experimental result demonstrated that our approach has a potential usefulness to assess the nodule evolution using thoracic CT images.

Pulmonary nodules are classified into three types such as solid, mixed GGO, and pure GGO types on the basis of the visual assessment of CT appearance. In our current study a quantitative classification algorithm has been developed by using volumetric data sets obtained from thin-section CT images. The algorithm can classify the pulmonary nodules into five types (α, β, γ, δ, and ε) on the basis of internal features extracted from CT number histograms inside nodules. We applied dynamic enhanced single slice and multi slice CT images to this classification algorithm and we analyzed it in each type.

Low radiation dose requirements create relatively noisy images that contribute to high numbers of false positive detections in CAD for CT colonography. Presumably image denoising techniques such as non-linear, edge-preserving smoothing filters can improve automatic colonic polyp detection in CT colonography by reducing overall per patient false positive rates. Here, we have evaluated multiple edge-preserving smoothing filters to determine whether this is so. Prone and supine scans from 81 asymptomatic, average-risk adults with adenomatous polyps were studied with and without smoothing. FROC curves were generated to analyze CAD results. A single, clinically relevant operating point was compared between the best smoothing filter results and the unsmoothed data. Improvement in performance was observed, but the differences were not found to be statistically significant for average dose CT colonography.

The computed tomographic colonography (CTC) computer aided detection (CAD) program is a new method in development to detect colon polyps in virtual colonoscopy. While high sensitivity is consistently achieved, additional features are desired to increase specificity. In this paper, a wavelet analysis was applied to CTCCAD outputs in an attempt to filter out false positive detections. 52 CTCCAD detection images were obtained using a screen capture application. 26 of these images were real polyps, confirmed by optical colonoscopy and 26 were false positive detections. A discrete wavelet transform of each image was computed with the MATLAB wavelet toolbox using the Haar wavelet at levels 1-5 in the horizontal, vertical and diagonal directions. From the resulting wavelet coefficients at levels 1-3 for all directions, a 72 feature vector was obtained for each image, consisting of descriptive statistics such as mean, variance, skew, and kurtosis at each level and orientation, as well as error statistics based on a linear predictor of neighboring wavelet coefficients. The vectors for each of the 52 images were then run through a support vector machine (SVM) classifier using ten-fold cross-validation training to determine its efficiency in distinguishing polyps from false positives. The SVM results showed 100% sensitivity and 51% specificity in correctly identifying the status of detections. If this technique were added to the filtering process of the CTCCAD polyp detection scheme, the number of false positive results could be reduced significantly.

Patient motion during SPECT acquisition causes inconsistent projection data and reconstruction artifacts which can significantly affect the diagnostic accuracy of SPECT. The tracking of motion by infrared monitoring spherical reflectors (markers) on the patient's surface can provide 6-Degrees-of-Freedom (6-DOF) motion information capable of providing clinically robust correction. Object rigid-body motion can be described by 3 translational DOF and 3 rotational DOF. Polaris marker position information obtained by stereo infrared cameras requires algorithmic processing to correctly record the tracked markers, and to calibrate and map Polaris co-ordinate data into the SPECT co-ordinate system. Marker data then requires processing to determine the rotational and translational 6-DOF motion to ultimately be used for SPECT image corrections. This processing utilizes an algorithm involving least-squares fitting, to each other, of two 3-D point sets using singular value decomposition (SVD) resulting in the rotation matrix and translation of the rigid body centroid. We have demonstrated the ability to monitor 12 clinical patients as well as 7 markers on 2 elastic belts worn by a volunteer while intentionally moving, and determined the 3 axis Euclidian rotation angles and centroid translations. An anthropomorphic phantom with Tc-99m added to the heart, liver, and body was simultaneously SPECT imaged and motion tracked using 4 rigidly mounted markers. The determined rotation matrix and translation information was used to correct the image resulting in virtually identical "no motion" and "corrected" images. We plan to initiate routine 6-DOF tracking of patient motion during SPECT imaging in the future.

We have developed a software, which allows to do non conventional percent quantitative analysis on scintigraphic polar map obtained from conventional processing of gated-SPECT acquisitions. Polar maps are 8 bit images of perfusion, motion, ejection fraction (EF) and thickening, of the heart. The software is written in Matlab, analyses the whole polar map and four ROIs corresponding to the theoretical LAD, LCX, RCA territories (perfused by these arteries) and extra-ROIs region. An intensity segmentation is performed. The area corresponding to pixels lower and higher than a varying cut-off are calculated on the whole image and for each ROI. The software calculates an intensity-area histogram, which is the analogous of the Dose-Volume Histogram used in radiation therapy: in this case, the histogram has the meaning of a Perfusion- or a Motion-Volume histogram. Then, the software applies the Lyman-Wolbarst algorithm, to calculate the area equivalent histogram reduction (e.g. the perfused area in the hypothesis that all pixels are perfused at 100%.). The makes a direct comparison between two different polar maps by choice. The comparison between the numerical quantification of motion and perfusion maps, allows the physicians to get a clinical evaluation of the stunned myocardium.

Electrical Impedance Tomography (EIT) seeks to recover the impedance distribution within a body using boundary data. More specifically, given the measured potentials, the model of the body - an elliptic partial differential equation - and the boundary conditions, this technique solves a non-linear inverse problem for the unknown impedance. In this work, an algorithm called Topology Optimization Method (TOM) is applied to EIT and compared to the Gauss-Newton Method (GNM). The Topology Optimization has solved some non-linear inverse problems and some of its procedures were not investigated for EIT, for instance, the use of Sequential Linear Programming. Assuming a pure resistive medium, the static resistivity distribution of a phantom was estimated using a 2-D finite element model. While the first method (GNM) essentially solves several algebraic systems, the second (TOM) solves several linear programming problems. Results using experimental data are shown and the quality of the images obtained, time and memory used are compared for both algorithms. We intend to use these methods, in future works, for the visualization of a human lung subjected to mechanical ventilation.

In many parts of the world, breast cancer is the leading cause mortality among women and it is the major cause of cancer death, next only to lung cancer. In recent years, microwave imaging has shown its potential as an alternative approach for breast cancer detection. Although advances have improved the likelihood of developing an early detection system based on this technology, there are still limitations. One of these limitations is that target responses are often obscured by surface reflections. Contrary to ground penetrating radar applications, a simple reference subtraction cannot be easily applied to alleviate this problem due to differences in the breast skin composition between patients. A novel surface removal technique for the removal of these high intensity reflections is proposed in this paper. This paper presents an algorithm based on the multiplication of adjacent wavelet subbands in order to enhance target echoes while reducing skin reflections. In these multiscale products, target signatures can be effectively distinguished from surface reflections. A simple threshold is applied to the signal in the wavelet domain in order to eliminate the skin responses. This final signal is reconstructed to the spatial domain in order to obtain a focused image. The proposed algorithm yielded promising results when applied to real data obtained from a phantom which mimics the dielectric properties of breast, cancer and skin tissues.

Early detection of glaucoma is essential to minimizing the risk of visual loss. It has been shown that a good predictor of glaucoma is the cup-to-disc ratio of the optic nerve head. This paper presents a highly automated method to segment the 'rim' (disc) and 'cup' from the optic nerve head in stereo images and calculate the cup-to-disc ratio. In this approach, the optic nerve head is unwrapped in polar coordinates and represented as a graph. Utilizing a novel and efficient graph searching technique for determining globally optimal closed-paths and an intelligent cost function, the rim and the cup are segmented from the stereo images. The results offer a more intuitive quantitative analysis compared to current planimetry-based techniques because the ophthalmologist can view the segmented images along with the derived cup-to-disc ratio.

The concept of the biopsy is ubiquitous in current medical diagnosis of cancer and other diseases. The standard biopsy consists of removing a sample of tissue for evaluation and diagnosis, primarily to ascertain the presence of cancer cells by (histo)pathological analyses. However, the advent of new optical imaging modalities and targeted or "smart" agents, that have affinity for a select target, suggests the possibility of performing in vivo tissue characterization without the need for sample removal or the wait for histopathologic processing. Here we present work testing and validating a fiber-based confocal fluorescence microscopic imaging system intended for combination with a larger scale imaging modality (i.e. MRI or CT) to be used in image-guided in vivo tissue characterization. Fiber-based confocal fluorescence microscopic imaging experiments were performed (Cellvizio, Mauna Kea Technologies, Paris, France) in vivo in two mouse models including: 1) EGFP-expressing mouse melanoma model and 2) M21 mouse melanoma model. Both models are known to express integrin α_νβ₃, a cell-surface receptor protein. We also performed an experiment in ex vivo chicken muscle tissue labelled with a fluorescein isothiocyanate-lectin targeted compound. In the mouse models, contrast agents that targeted the integrin were injected and the contrast agent localization in tumor was verified by a whole-body multispectral imager. The fiber-based tool was sensitive enough to detect and image the tissue of interest in all different experiments, and was found appropriate for use in interventional catheter-based procedures.

Expression of human Aβ42 peptide in the Drosophila brain induces pathological phenotypes resembling Alzheimer's disease (PNAS 101, 6623-6628). Three-dimensional confocal imaging reveals extensive vacuoles caused by neurodegeneration in the brain of aged but not young Aβ42 flies. Here, we report a three-dimensional computation algorism allowing semi-automatic measurement of numbers and volumes of brain vacuoles. The method employed matched filters, α-shape, and the active-contour techniques. Using this method, a good result depicting the contours of the vacuoles can be obtained. A more accurate algorithm is still under development. Accurate evaluation of brain pathology in Alzheimer's flies may facilitate the understanding of molecular mechanisms underlying Aβ toxicity and the discovery of novel therapeutic targets for Alzheimer's disease.

The purpose of this study is to demonstrate the feasibility of performing cone beam CT angiogenesis imaging (CBCTAI) using a cone beam CT scanner prototype. We have developed a cone beam breast CT imaging technique that is able to detect a ~2 mm tumor in live mice and have expanded the imaging to include visualization of tumor vessels. The CBCT prototype consisted of a modified clinical CT scanner and a flat panel detector. This CBCT scanner prototype was used for a series of preliminary contrast studies with live mice: contrast-to-scan time delay, contrast for tumor vessel enhancement (comparison with histology) and early tumor vessel development imaging. The results of the live mice studies demonstrate that good image quality can be achieved with this prototype demonstrating the feasibility of CBCTAI. Achieving the CBCTAI technique on the cone beam breast CT imaging modality will significantly advance breast cancer detection, diagnosis and treatment.

Repeated micro-CT scanning of a number of iliac crest biopsies enabled us to quantitate the variation in CT image gray-scale and spatial geometry due to variables such as specimen orientation, projection magnification, voxel size and slight differences in x-ray photon energy in each of the different scans. Using the micro-CT scanner on beamline X2B at the Brookhaven National Laboratory's National Synchrotron Light Source, we rescanned several iliac crest bone biopsy specimens, and a test phantom made of calcium hydroxyapatite, at repeated scanning sessions and evaluated the reproducibility of the spatial geometry and gray-scale haracteristics of the specimens. This scanner consists of a Bragg diffiaction source of monochromatic x-rays, a computer controlled high precision specimen rotation and translation stage assembly, and a fluorescent crystal and CCD array system for imaging the specimen at each of the angles of view around its axis of rotation during the scanning sequence. The 3-D micro-CT images consisted of up to 1024x2400², 4 μm, cubic voxels, each with 16-bit gray-scale. We also reconstructed the images at 16,32 and 48 μm voxel resolution. Partial volume effects at the surface of the bone were diminished by 'eroding' the surface voxels in the 4 μm images, but significantly changed the outcome at greater voxel size. Reproducibility of the mineral content of bone, at mean bone opacity value, was ± 28.8 mg/cm³, i.e., 2.56%, in a 4 μm cubic voxel at the 95% confidence level.

We are investigating imaging techniques to study the tumor response to photodynamic therapy (PDT). PET can provide physiological and functional information. High-resolution MRI can provide anatomical and morphological changes. Image registration can combine MRI and PET images for improved tumor monitoring. In this study, we acquired high-resolution MRI and microPET [¹⁸F]fluorodeoxyglucose (FDG) images from C3H mice with RIF-1 tumors that were treated with Pc 4-based PDT. For tumor registration, we developed a finite element model (FEM)-based deformable registration scheme. To assess the registration quality, we performed slice by slice review of both image volumes, computed the volume overlap ratios, and visualized both volumes in color overlay. The mean volume overlap ratios for tumors were 94.7% after registration. Registration of high-resolution MRI and microPET images combines anatomical and functional information of the tumors and provides a useful tool for evaluating photodynamic therapy.