Vahid Ghodrati1,2, Jiaxin Shao1, Mark Bydder1, Ziwu Zhou1,3, Wotao Yin4, Kim-Lien Nguyen5, Yingli Yang2,6, Peng Hu1,2. 1. Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, CA, USA. 2. Biomedical Physics Inter-Departmental Graduate Program, University of California, Los Angeles, CA, USA. 3. Department of Bioengineering, University of California, Los Angeles, CA, USA. 4. Department of Mathematics, University of California, Los Angeles, CA, USA. 5. Division of Cardiology, Department of Medicine, David Geffen School of Medicine, University of California, Los Angeles, CA, USA. 6. Department of Radiation Oncology, University of California, Los Angeles, CA, USA.
Abstract
BACKGROUND: To review and evaluate approaches to convolutional neural network (CNN) reconstruction for accelerated cardiac MR imaging in the real clinical context. METHODS: Two CNN architectures, Unet and residual network (Resnet) were evaluated using quantitative and qualitative assessment by radiologist. Four different loss functions were also considered: pixel-wise (L1 and L2), patch-wise structural dissimilarity (Dssim) and feature-wise (perceptual loss). The networks were evaluated using retrospectively and prospectively under-sampled cardiac MR data. RESULTS: Based on our assessments, we find that Resnet and Unet achieve similar image quality but that former requires only 100,000 parameters compared to 1.3 million parameters for the latter. The perceptual loss function performed significantly better than L1, L2 or Dssim loss functions as determined by the radiologist scores. CONCLUSIONS: CNN image reconstruction using Resnet yields comparable image quality to Unet with 10X the number of parameters. This has implications for training with significantly lower data requirements. Network training using the perceptual loss function was found to better agree with radiologist scoring compared to L1, L2 or Dssim loss functions. 2019 Quantitative Imaging in Medicine and Surgery. All rights reserved.
BACKGROUND: To review and evaluate approaches to convolutional neural network (CNN) reconstruction for accelerated cardiac MR imaging in the real clinical context. METHODS: Two CNN architectures, Unet and residual network (Resnet) were evaluated using quantitative and qualitative assessment by radiologist. Four different loss functions were also considered: pixel-wise (L1 and L2), patch-wise structural dissimilarity (Dssim) and feature-wise (perceptual loss). The networks were evaluated using retrospectively and prospectively under-sampled cardiac MR data. RESULTS: Based on our assessments, we find that Resnet and Unet achieve similar image quality but that former requires only 100,000 parameters compared to 1.3 million parameters for the latter. The perceptual loss function performed significantly better than L1, L2 or Dssim loss functions as determined by the radiologist scores. CONCLUSIONS: CNN image reconstruction using Resnet yields comparable image quality to Unet with 10X the number of parameters. This has implications for training with significantly lower data requirements. Network training using the perceptual loss function was found to better agree with radiologist scoring compared to L1, L2 or Dssim loss functions. 2019 Quantitative Imaging in Medicine and Surgery. All rights reserved.
Entities:
Keywords:
Magnetic resonance imaging; cardiac image reconstruction; convolutional Unet; deep learning; perceptual loss function; residual neural network
Authors: Mark A Griswold; Peter M Jakob; Robin M Heidemann; Mathias Nittka; Vladimir Jellus; Jianmin Wang; Berthold Kiefer; Axel Haase Journal: Magn Reson Med Date: 2002-06 Impact factor: 4.668
Authors: Hu Chen; Yi Zhang; Mannudeep K Kalra; Feng Lin; Yang Chen; Peixi Liao; Jiliu Zhou; Ge Wang Journal: IEEE Trans Med Imaging Date: 2017-06-13 Impact factor: 10.048
Authors: Jo Schlemper; Jose Caballero; Joseph V Hajnal; Anthony N Price; Daniel Rueckert Journal: IEEE Trans Med Imaging Date: 2017-10-13 Impact factor: 10.048
Authors: Kerstin Hammernik; Teresa Klatzer; Erich Kobler; Michael P Recht; Daniel K Sodickson; Thomas Pock; Florian Knoll Journal: Magn Reson Med Date: 2017-11-08 Impact factor: 4.668
Authors: Tianming Du; Honggang Zhang; Yuemeng Li; Stephen Pickup; Mark Rosen; Rong Zhou; Hee Kwon Song; Yong Fan Journal: Med Image Anal Date: 2021-05-16 Impact factor: 13.828
Authors: Akshay S Chaudhari; Christopher M Sandino; Elizabeth K Cole; David B Larson; Garry E Gold; Shreyas S Vasanawala; Matthew P Lungren; Brian A Hargreaves; Curtis P Langlotz Journal: J Magn Reson Imaging Date: 2020-08-24 Impact factor: 5.119