Seonyeong Park1, H Michael Gach2, Siyong Kim3, Suk Jin Lee4, Yuichi Motai5. 1. Department of BioengineeringUniversity of Illinois at Urbana-Champaign Urbana IL 61820 USA. 2. Department of Radiation OncologyWashington University in St. Louis St. Louis MO 63130 USA. 3. Department of Radiation OncologyDivision of Medical PhysicsVirginia Commonwealth University Richmond VA 23284 USA. 4. TSYS School of Computer ScienceColumbus State University Columbus GA 31907 USA. 5. Department of Electrical and Computer EngineeringVirginia Commonwealth University Richmond VA 23284 USA.
Abstract
OBJECTIVE: To introduce an MRI in-plane resolution enhancement method that estimates High-Resolution (HR) MRIs from Low-Resolution (LR) MRIs. METHOD & MATERIALS: Previous CNN-based MRI super-resolution methods cause loss of input image information due to the pooling layer. An Autoencoder-inspired Convolutional Network-based Super-resolution (ACNS) method was developed with the deconvolution layer that extrapolates the missing spatial information by the convolutional neural network-based nonlinear mapping between LR and HR features of MRI. Simulation experiments were conducted with virtual phantom images and thoracic MRIs from four volunteers. The Peak Signal-to-Noise Ratio (PSNR), Structure SIMilarity index (SSIM), Information Fidelity Criterion (IFC), and computational time were compared among: ACNS; Super-Resolution Convolutional Neural Network (SRCNN); Fast Super-Resolution Convolutional Neural Network (FSRCNN); Deeply-Recursive Convolutional Network (DRCN). RESULTS: ACNS achieved comparable PSNR, SSIM, and IFC results to SRCNN, FSRCNN, and DRCN. However, the average computation speed of ACNS was 6, 4, and 35 times faster than SRCNN, FSRCNN, and DRCN, respectively under the computer setup used with the actual average computation time of 0.15 s per [Formula: see text].
OBJECTIVE: To introduce an MRI in-plane resolution enhancement method that estimates High-Resolution (HR) MRIs from Low-Resolution (LR) MRIs. METHOD & MATERIALS: Previous CNN-based MRI super-resolution methods cause loss of input image information due to the pooling layer. An Autoencoder-inspired Convolutional Network-based Super-resolution (ACNS) method was developed with the deconvolution layer that extrapolates the missing spatial information by the convolutional neural network-based nonlinear mapping between LR and HR features of MRI. Simulation experiments were conducted with virtual phantom images and thoracic MRIs from four volunteers. The Peak Signal-to-Noise Ratio (PSNR), Structure SIMilarity index (SSIM), Information Fidelity Criterion (IFC), and computational time were compared among: ACNS; Super-Resolution Convolutional Neural Network (SRCNN); Fast Super-Resolution Convolutional Neural Network (FSRCNN); Deeply-Recursive Convolutional Network (DRCN). RESULTS: ACNS achieved comparable PSNR, SSIM, and IFC results to SRCNN, FSRCNN, and DRCN. However, the average computation speed of ACNS was 6, 4, and 35 times faster than SRCNN, FSRCNN, and DRCN, respectively under the computer setup used with the actual average computation time of 0.15 s per [Formula: see text].
Entities:
Keywords:
Autoencoder; MRI; convolution neural network; deep learning; super resolution
Authors: Can Zhao; Muhan Shao; Aaron Carass; Hao Li; Blake E Dewey; Lotta M Ellingsen; Jonghye Woo; Michael A Guttman; Ari M Blitz; Maureen Stone; Peter A Calabresi; Henry Halperin; Jerry L Prince Journal: Magn Reson Imaging Date: 2019-06-24 Impact factor: 2.546
Authors: Qing Lyu; Hongming Shan; Cole Steber; Corbin Helis; Chris Whitlow; Michael Chan; Ge Wang Journal: IEEE Trans Med Imaging Date: 2020-02-18 Impact factor: 10.048
Authors: Shuzhou Jiang; Hui Xue; Alan Glover; Mary Rutherford; Daniel Rueckert; Joseph V Hajnal Journal: IEEE Trans Med Imaging Date: 2007-07 Impact factor: 10.048