Development of a novel algorithm to improve image quality in chest digital tomosynthesis using convolutional neural network with super-resolution

Tsutomu Gomi

doi:10.1117/12.2581917

15 February 2021 Development of a novel algorithm to improve image quality in chest digital tomosynthesis using convolutional neural network with super-resolution

Tsutomu Gomi

Author Affiliations +

Proceedings Volume 11595, Medical Imaging 2021: Physics of Medical Imaging; 115953F (2021) https://doi.org/10.1117/12.2581917
Event: SPIE Medical Imaging, 2021, Online Only

Conference Poster

Abstract

We developed a novel dual-energy (DE) virtual monochromatic (VM) very-deep super-resolution (VDSR) reconstruction algorithm (DVV) that uses projection data to improve nodule contrast and resolution during chest digital tomosynthesis (CDT). To estimate residual errors in high-resolution and multiscale VM images in projection space, the DVV algorithm employs a training network (mini-batch stochastic gradient descent algorithm with momentum) that involves subjectively reconstructed hybrid SR images [simultaneous algebraic reconstruction technique (SART) total variation (TV)-first iterative shrinkage-thresholding algorithm (FISTA); SART-TV-FISTA]. DE-DT imaging was accomplished using pulsed X-ray exposures that were rapidly switched between low- and high-tube potential kVp, followed by image comparisons using conventional polychromatic filtered back projection (FBP), SART-TV-FISTA, and DE-VM-SART-TV-FISTA algorithms. Improvements in contrast and resolution were compared using signal difference- to-noise ratio (SDNR) and radial-modulation transfer function (radial-MTF) of a chest phantom with simulated ground-glass opacity (GGO) nodules. The novel DVV algorithm improved overall performance in terms of SDNR and yielded high quality images independent of the type of simulated GGO nodules used in the chest phantom. The novel DVV algorithm yielded superior resolution in comparison with conventional reconstruction algorithms for the radial-MTF analysis, with and without VM processing. Furthermore, the DVV algorithm improved contrast and spatial resolution.

Conference Presentation

Citation Download Citation

Tsutomu Gomi "Development of a novel algorithm to improve image quality in chest digital tomosynthesis using convolutional neural network with super-resolution", Proc. SPIE 11595, Medical Imaging 2021: Physics of Medical Imaging, 115953F (15 February 2021); https://doi.org/10.1117/12.2581917

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

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.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available