Open AccessHigher‐resolution quantification of white matter hypointensities by large‐scale transfer learning from 2D images on the JPSC‐AD cohort
Author(s) -
Thyreau Benjamin,
Tatewaki Yasuko,
Chen Liying,
Takano Yuji,
Hirabayashi Naoki,
Furuta Yoshihiko,
Hata Jun,
Nakaji Shigeyuki,
Maeda Tetsuya,
NoguchiShinohara Moeko,
Mimura Masaru,
Nakashima Kenji,
Mori Takaaki,
Takebayashi Minoru,
Ninomiya Toshiharu,
Taki Yasuyuki
Publication year - 2022
Publication title -
human brain mapping
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.005
H-Index - 191
eISSN - 1097-0193
pISSN - 1065-9471
DOI - 10.1002/hbm.25899
Subject(s) - fluid attenuated inversion recovery , convolutional neural network , computer science , artificial intelligence , deep learning , pattern recognition (psychology) , segmentation , transfer of learning , hyperintensity , white matter , machine learning , magnetic resonance imaging , medicine , radiology
White matter lesions (WML) commonly occur in older brains and are quantifiable on MRI, often used as a biomarker in Aging research. Although algorithms are regularly proposed that identify these lesions from T2-fluid-attenuated inversion recovery (FLAIR) sequences, none so far can estimate lesions directly from T1-weighted images with acceptable accuracy. Since 3D T1 is a polyvalent and higher-resolution sequence, it could be beneficial to obtain the distribution of WML directly from it. However a serious difficulty, both for algorithms and human, can be found in the ambiguities of brain signal intensity in T1 images. This manuscript shows that a cross-domain ConvNet (Convolutional Neural Network) approach can help solve this problem. Still, this is non-trivial, as it would appear to require a large and varied dataset (for robustness) labelled at the same high resolution (for spatial accuracy). Instead, our model was taught from two-dimensional FLAIR images with a loss function designed to handle the super-resolution need. And crucially, we leveraged a very large training set for this task, the recently assembled, multi-sites Japan Prospective Studies Collaboration for Aging and Dementia (JPSC-AD) cohort. We describe the two-step procedure that we followed to handle such a large number of imperfectly labeled samples. A large-scale accuracy evaluation conducted against FreeSurfer 7, and a further visual expert rating revealed that WML segmentation from our ConvNet was consistently better. Finally, we made a directly usable software program based on that trained ConvNet model, available at https://github.com/bthyreau/deep-T1-WMH.