Self-supervised learning for spinal MRIs
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 sagi...
| Egile Nagusiak: | , , |
|---|---|
| Formatua: | Conference item |
| Argitaratua: |
Springer, Cham
2017
|
| _version_ | 1826284102434684928 |
|---|---|
| author | Jamaludin, A Kadir, T Zisserman, A |
| author_facet | Jamaludin, A Kadir, T Zisserman, A |
| author_sort | Jamaludin, A |
| collection | OXFORD |
| description | 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. |
| first_indexed | 2024-03-07T01:08:49Z |
| format | Conference item |
| id | oxford-uuid:8c469567-0741-4e32-996a-86222269b1a6 |
| institution | University of Oxford |
| last_indexed | 2024-03-07T01:08:49Z |
| publishDate | 2017 |
| publisher | Springer, Cham |
| record_format | dspace |
| spelling | oxford-uuid:8c469567-0741-4e32-996a-86222269b1a62022-03-26T22:43:34ZSelf-supervised learning for spinal MRIsConference itemhttp://purl.org/coar/resource_type/c_5794uuid:8c469567-0741-4e32-996a-86222269b1a6Symplectic Elements at OxfordSpringer, Cham2017Jamaludin, AKadir, TZisserman, AA 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. |
| spellingShingle | Jamaludin, A Kadir, T Zisserman, A Self-supervised learning for spinal MRIs |
| title | Self-supervised learning for spinal MRIs |
| title_full | Self-supervised learning for spinal MRIs |
| title_fullStr | Self-supervised learning for spinal MRIs |
| title_full_unstemmed | Self-supervised learning for spinal MRIs |
| title_short | Self-supervised learning for spinal MRIs |
| title_sort | self supervised learning for spinal mris |
| work_keys_str_mv | AT jamaludina selfsupervisedlearningforspinalmris AT kadirt selfsupervisedlearningforspinalmris AT zissermana selfsupervisedlearningforspinalmris |