SpineNet: Automated classification and evidence visualization in spinal MRIs
The objective of this work is to automatically produce radiological gradings of spinal lumbar MRIs and also localize the predicted pathologies. We show that this can be achieved via a Convolutional Neural Network (CNN) framework that takes intervertebral disc volumes as inputs and is trained only on...
| Główni autorzy: | , , |
|---|---|
| Format: | Journal article |
| Język: | English |
| Wydane: |
Elsevier
2017
|
| _version_ | 1826277702290636800 |
|---|---|
| author | Jamaludin, A Kadir, T Zisserman, A |
| author_facet | Jamaludin, A Kadir, T Zisserman, A |
| author_sort | Jamaludin, A |
| collection | OXFORD |
| description | The objective of this work is to automatically produce radiological gradings of spinal lumbar MRIs and also localize the predicted pathologies. We show that this can be achieved via a Convolutional Neural Network (CNN) framework that takes intervertebral disc volumes as inputs and is trained only on disc-specific class labels. Our contributions are: (i) a CNN architecture that predicts multiple gradings at once, and we propose variants of the architecture including using 3D convolutions; (ii) showing that this architecture can be trained using a multi-task loss function without requiring segmentation level annotation; and (iii) a localization method that clearly shows pathological regions in the disc volumes. We compare three visualization methods for the localization. The network is applied to a large corpus of MRI T2 sagittal spinal MRIs (using a standard clinical scan protocol) acquired from multiple machines, and is used to automatically compute disk and vertebra gradings for each MRI. These are: Pfirrmann grading, disc narrowing, upper/lower endplate defects, upper/lower marrow changes, spondylolisthesis, and central canal stenosis. We report near human performances across the eight gradings, and also visualize the evidence for these gradings localized on the original scans. |
| first_indexed | 2024-03-06T23:32:54Z |
| format | Journal article |
| id | oxford-uuid:6caacb46-d262-4902-92cd-b9c30aee8520 |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-03-06T23:32:54Z |
| publishDate | 2017 |
| publisher | Elsevier |
| record_format | dspace |
| spelling | oxford-uuid:6caacb46-d262-4902-92cd-b9c30aee85202022-03-26T19:12:31ZSpineNet: Automated classification and evidence visualization in spinal MRIsJournal articlehttp://purl.org/coar/resource_type/c_dcae04bcuuid:6caacb46-d262-4902-92cd-b9c30aee8520EnglishSymplectic Elements at OxfordElsevier2017Jamaludin, AKadir, TZisserman, AThe objective of this work is to automatically produce radiological gradings of spinal lumbar MRIs and also localize the predicted pathologies. We show that this can be achieved via a Convolutional Neural Network (CNN) framework that takes intervertebral disc volumes as inputs and is trained only on disc-specific class labels. Our contributions are: (i) a CNN architecture that predicts multiple gradings at once, and we propose variants of the architecture including using 3D convolutions; (ii) showing that this architecture can be trained using a multi-task loss function without requiring segmentation level annotation; and (iii) a localization method that clearly shows pathological regions in the disc volumes. We compare three visualization methods for the localization. The network is applied to a large corpus of MRI T2 sagittal spinal MRIs (using a standard clinical scan protocol) acquired from multiple machines, and is used to automatically compute disk and vertebra gradings for each MRI. These are: Pfirrmann grading, disc narrowing, upper/lower endplate defects, upper/lower marrow changes, spondylolisthesis, and central canal stenosis. We report near human performances across the eight gradings, and also visualize the evidence for these gradings localized on the original scans. |
| spellingShingle | Jamaludin, A Kadir, T Zisserman, A SpineNet: Automated classification and evidence visualization in spinal MRIs |
| title | SpineNet: Automated classification and evidence visualization in spinal MRIs |
| title_full | SpineNet: Automated classification and evidence visualization in spinal MRIs |
| title_fullStr | SpineNet: Automated classification and evidence visualization in spinal MRIs |
| title_full_unstemmed | SpineNet: Automated classification and evidence visualization in spinal MRIs |
| title_short | SpineNet: Automated classification and evidence visualization in spinal MRIs |
| title_sort | spinenet automated classification and evidence visualization in spinal mris |
| work_keys_str_mv | AT jamaludina spinenetautomatedclassificationandevidencevisualizationinspinalmris AT kadirt spinenetautomatedclassificationandevidencevisualizationinspinalmris AT zissermana spinenetautomatedclassificationandevidencevisualizationinspinalmris |