Lung opacities on CT scans, such as ground-glass opacities (GGOs) and consolidation, can manifest with various conditions, including lung cancers, pulmonary edema and COVID-19. Presentation of these non-specific findings can vary from isolated focal to diffuse opacities in all lobes. Moreover, disease distributions and progressions vary across disease types and patients. This unpredictability can challenge one’s ability to accurately quantify and compare the percentage of infected lung within and across patients. Despite the promise of AI models for image segmentation, the inconsistency of lung opacities, and limited access to large annotated datasets affect generalization performance of models to the cohort of lung diseases. In this paper, we developed a single-stage system to jointly localize the lung and opacifications in CT scans using a diverse real-world dataset with sparse annotations. A multi-class Dense U-Net model was designed to segment the lungs and two classes of opacity regions (GGOs and consolidations) in CT images. The model was trained on 4075 slices from 495 sparsely annotated CT studies and evaluated on 18625 slices from 103 densely annotated studies (37 positive). A comparative analysis of different training data subsets and loss functions was performed to determine optimal model design. Performance was evaluated by comparing manual and automated lung opacity percentages via Pearson Correlation Coefficient. The optimal model achieved a Pearson Correlation Coefficient of 0.99. These findings suggest the potential of developing an accurate method to localize lung opacification, unspecific to a particular disease
In this study, we propose and validate an end-to-end pipeline based on deep learning for differential diagnosis of emphysema in thoracic CT images. The five lung tissue patterns involved in most differential restrictive and obstructive lung disease diagnoses include: emphysema, ground glass, fibrosis, micronodule, and normal. Four established network architectures have been trained and evaluated. To the best of our knowledge, this is the first comprehensive end-to-end deep CNN pipeline for differential diagnosis of emphysema. A comparative analysis shows the performance of the proposed models on two publicly available datasets.
KEYWORDS: Image segmentation, Lung, Computed tomography, 3D modeling, Convolutional neural networks, Medical imaging, 3D image processing, Data modeling, Image processing
Deep Learning models such as Convolutional Neural Networks (CNNs) have achieved state-of-the-art performance in 2D medical image analysis. In clinical practice; however, most analyzed and acquired medical data are formed of 3D volumes. In this paper, we present a fast and efficient 3D lung segmentation method based on V-net: a purely volumetric fully CNN. Our model is trained on chest CT images through volume to volume learning, which palliates overfitting problem on limited number of annotated training data. Adopting a pre-processing step and training an objective function based on Dice coefficient addresses the imbalance between the number of lung voxels against that of background. We have leveraged Vnet model by using batch normalization for training which enables us to use higher learning rate and accelerates the training of the model. To address the inadequacy of training data and obtain better robustness, we augment the data applying random linear and non-linear transformations. Experimental results on two challenging medical image data show that our proposed method achieved competitive result with a much faster speed.
Flow Doppler imaging is widely used by clinicians to detect diseases of the valves. In particular, continuous wave (CW) Doppler mode scan is routinely done during echocardiography and shows Doppler signal traces over multiple heart cycles. Traditionally, echocardiographers have manually traced such velocity envelopes to extract measurements such as decay time and pressure gradient which are then matched to normal and abnormal values based on clinical guidelines. In this paper, we present a fully automatic approach to deriving these measurements for aortic stenosis retrospectively from echocardiography videos. Comparison of our method with measurements made by echocardiographers shows large agreement as well as identification of new cases missed by echocardiographers.
Since many diseases or injuries can cause biomechanical or structural property changes that can alter lung function, there is great interest in measuring regional lung function by measurement of regional mechanical changes. To date, the most prevalent approach for assessing regional lung function from 4-D X-ray CT data has been a measure of Jacobian of deformation. However, although the Jacobian describes regional volume changes of the lung during deformation, it lacks any consideration of directional changes of local compressions and expansions during respiration. Herein, we propose the use of strain as a measure of regional lung function from 4-D thoracic CT and we perform correlation of principal strains of calculated deformation by s recently proposed 3-D optical flow technique (MOFID) computed from radiotherapy treatment planning 4-D X-ray CT data sets collected in seven subjects suffering from non-small cell primary lung cancer. In addition to 4-D CT data, both SPECT ventilation (VSPECT), and SPECT perfusion (QSPECT) data were acquired in all subjects. For each subject, we performed voxel-wise statistical correlation of the Jacobian as well as principal strains of deformation (CT-derived pulmonary function images) with both ventilation and perfusion SPECT. For all subjects, the maximum principal strain resulted in a higher correlation with both SPECT ventilation and SPECT perfusion than other indices including the previously established Jacobian metric.
The accuracy of optical flow estimation algorithms has been improving steadily by refining the objective function which
should be optimized. A novel energy function for computing optical flow from volumetric X-ray CT images is presented.
One advantage of the optical flow framework is the possibility to enforce physical constraints on the numerical solutions.
The physical constraints which have been included here are: brightness constancy, gradient constancy, continuity
equation based on mass conservation, and discontinuity-preserving spatio-temporal smoothness. The method has been
evaluated on POPI-model and the evaluation demonstrates that the method results in significantly better accuracy than
previous optical flow techniques for estimation of deformable lung motion.
The treatment plan evaluation for lung cancer patients involves pre-treatment and post-treatment volume CT imaging of
the lung. However, treatment of the tumor volume lung results in structural changes to the lung during the course of
treatment. In order to register the pre-treatment volume to post-treatment volume, there is a need to find robust and
homologous features which are not affected by the radiation treatment along with a smooth deformation field. Since
airways are well-distributed in the entire lung, in this paper, we propose use of airway tree bifurcations for registration of
the pre-treatment volume to the post-treatment volume. A dedicated and automated algorithm has been developed that
finds corresponding airway bifurcations in both images. To derive the 3-D deformation field, a B-spline transformation
model guided by mutual information similarity metric was used to guarantee the smoothness of the transformation while
combining global information from bifurcation points. Therefore, the approach combines both global statistical intensity
information with local image feature information. Since during normal breathing, the lung undergoes large nonlinear
deformations, it is expected that the proposed method would also be applicable to large deformation registration between
maximum inhale and maximum exhale images in the same subject. The method has been evaluated by registering 3-D
CT volumes at maximum exhale data to all the other temporal volumes in the POPI-model data.
For a common set of 4D CT lung images, we report results from the application of a number of optical flow techniques
including global, local, and combined local-global methods for tracking planar lung motion. Our comparisons are
primarily empirical, and concentrate on the accuracy, that is magnitude and direction of error at a discrete set of
landmark points with known motions, and reliability, that is, objective lung boundary deformation tracking. Our results
indicate that performance varies significantly among the techniques tested.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.