Self-supervised learning for spinal MRIs

Conference item

A significant proportion of patients scanned in a clinical setting have follow-up scans. We show in this work that such longitudinal scans alone can be used as a form of “free” self-supervision for training a deep network. We demonstrate this self-supervised learning for the case of T2-weighted sagittal lumbar Magnetic Resonance Images (MRIs). A Siamese convolutional neural network (CNN) is trained using two losses: (i) a contrastive loss on whether the scan is of the same person (i.e. longitudinal) or not, together with (ii) a classification loss on predicting the level of vertebral bodies. The performance of this pre-trained network is then assessed on a grading classification task. We experiment on a dataset of 1016 subjects, 423 possessing follow-up scans, with the end goal of learning the disc degeneration radiological gradings attached to the intervertebral discs. We show that the performance of the pre-trained CNN on the supervised classification task is (i) superior to that of a network trained from scratch; and (ii) requires far fewer annotated training samples to reach an equivalent performance to that of the network trained from scratch.

Authors: Jamaludin, A; Kadir, T; Zisserman, A

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: Springer, Cham
• Host title: DLMIA 2017, ML-CDS 2017: Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support
• Journal: DLMIA 2017, ML-CDS 2017
• Volume: 10553
• Pages: 294-302
• Series: Lecture Notes in Computer Science
• Publication date: 2017-09-09
• Acceptance date: 2017-07-10
• DOI: 10.1007/978-3-319-67558-9_34
• ISSN: 0302-9743
• ISBN: 9783319675589