Title | Deep neural network for water/fat separation: Supervised training, unsupervised training, and no training. |
Publication Type | Journal Article |
Year of Publication | 2021 |
Authors | Jafari R, Spincemaille P, Zhang J, Nguyen TD, Luo X, Cho J, Margolis D, Prince MR, Wang Y |
Journal | Magn Reson Med |
Volume | 85 |
Issue | 4 |
Pagination | 2263-2277 |
Date Published | 2021 04 |
ISSN | 1522-2594 |
Keywords | Algorithms, Deep Learning, Neural Networks, Computer, Water |
Abstract | PURPOSE: To use a deep neural network (DNN) for solving the optimization problem of water/fat separation and to compare supervised and unsupervised training. METHODS: The current -IDEAL algorithm for solving water/fat separation is dependent on initialization. Recently, DNN has been proposed to solve water/fat separation without the need for suitable initialization. However, this approach requires supervised training of DNN using the reference water/fat separation images. Here we propose 2 novel DNN water/fat separation methods: 1) unsupervised training of DNN (UTD) using the physical forward problem as the cost function during training, and 2) no training of DNN using physical cost and backpropagation to directly reconstruct a single dataset. The supervised training of DNN, unsupervised training of DNN, and no training of DNN methods were compared with the reference -IDEAL. RESULTS: All DNN methods generated consistent water/fat separation results that agreed well with -IDEAL under proper initialization. CONCLUSION: The water/fat separation problem can be solved using unsupervised deep neural networks. |
DOI | 10.1002/mrm.28546 |
Alternate Journal | Magn Reson Med |
PubMed ID | 33107127 |
PubMed Central ID | PMC7809709 |
Grant List | R01 CA181566 / CA / NCI NIH HHS / United States R01 NS095562 / NS / NINDS NIH HHS / United States R01 NS090464 / NS / NINDS NIH HHS / United States R01 DK116126 / DK / NIDDK NIH HHS / United States |
Related Institute:
MRI Research Institute (MRIRI)