Automated segmentation and prediction of intervertebral disc morphology and uniaxial deformations from MRI
Objective: The measurement of in vivo intervertebral disc (IVD) mechanics may be used to understand the etiology of IVD degeneration and low back pain (LBP). To this end, our lab has developed methods to measure IVD morphology and uniaxial compressive deformation (% change in IVD height) resulting f...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2023-09-01
|
| Series: | Osteoarthritis and Cartilage Open |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2665913123000456 |
| _version_ | 1827856106018308096 |
|---|---|
| author | James A. Coppock Nicole E. Zimmer Charles E. Spritzer Adam P. Goode Louis E. DeFrate |
| author_facet | James A. Coppock Nicole E. Zimmer Charles E. Spritzer Adam P. Goode Louis E. DeFrate |
| author_sort | James A. Coppock |
| collection | DOAJ |
| description | Objective: The measurement of in vivo intervertebral disc (IVD) mechanics may be used to understand the etiology of IVD degeneration and low back pain (LBP). To this end, our lab has developed methods to measure IVD morphology and uniaxial compressive deformation (% change in IVD height) resulting from dynamic activity, in vivo, using magnetic resonance images (MRI). However, due to the time-intensive nature of manual image segmentation, we sought to validate an image segmentation algorithm that could accurately and reliably reproduce models of in vivo tissue mechanics. Design: Therefore, we developed and evaluated two commonly employed deep learning architectures (2D and 3D U-Net) for the segmentation of IVDs from MRI. The performance of these models was evaluated for morphological accuracy by comparing predicted IVD segmentations (Dice similarity coefficient, mDSC; average surface distance, ASD) to manual (ground truth) measures. Likewise, functional reliability and precision were assessed by evaluating the intraclass correlation coefficient (ICC) and standard error of measurement (SEm) of predicted and manually derived deformation measures. Results: Peak model performance was obtained using the 3D U-net architecture, yielding a maximum mDSC = 0.9824 and component-wise ASDx = 0.0683 mm; ASDy = 0.0335 mm; ASDz = 0.0329 mm. Functional model performance demonstrated excellent reliability ICC = 0.926 and precision SEm = 0.42%. Conclusions: This study demonstrated that a deep learning framework can precisely and reliably automate measures of IVD function, drastically improving the throughput of these time-intensive methods. |
| first_indexed | 2024-03-12T12:19:55Z |
| format | Article |
| id | doaj.art-ae93c680c7d444c0801e6e4337e0f59f |
| institution | Directory Open Access Journal |
| issn | 2665-9131 |
| language | English |
| last_indexed | 2024-03-12T12:19:55Z |
| publishDate | 2023-09-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Osteoarthritis and Cartilage Open |
| spelling | doaj.art-ae93c680c7d444c0801e6e4337e0f59f2023-08-30T05:54:48ZengElsevierOsteoarthritis and Cartilage Open2665-91312023-09-0153100378Automated segmentation and prediction of intervertebral disc morphology and uniaxial deformations from MRIJames A. Coppock0Nicole E. Zimmer1Charles E. Spritzer2Adam P. Goode3Louis E. DeFrate4Department of Orthopedic Surgery, Duke University School of Medicine, United States; Department of Biomedical Engineering, Duke University, United StatesDepartment of Orthopedic Surgery, Duke University School of Medicine, United States; Department of Biomedical Engineering, Duke University, United StatesDepartment of Radiology, Duke University School of Medicine, United StatesDepartment of Orthopedic Surgery, Duke University School of Medicine, United States; Duke Clinical Research Institute, Duke University School of Medicine, United States; Department of Population Health Sciences, Duke University, United StatesDepartment of Orthopedic Surgery, Duke University School of Medicine, United States; Department of Biomedical Engineering, Duke University, United States; Department of Mechanical Engineering and Materials Science, Duke University, United States; Corresponding author. Duke University, Box 3093, Durham, NC, 27708, United States. Tel.: +(919) 681 9959; fax: +(919) 681 8490.Objective: The measurement of in vivo intervertebral disc (IVD) mechanics may be used to understand the etiology of IVD degeneration and low back pain (LBP). To this end, our lab has developed methods to measure IVD morphology and uniaxial compressive deformation (% change in IVD height) resulting from dynamic activity, in vivo, using magnetic resonance images (MRI). However, due to the time-intensive nature of manual image segmentation, we sought to validate an image segmentation algorithm that could accurately and reliably reproduce models of in vivo tissue mechanics. Design: Therefore, we developed and evaluated two commonly employed deep learning architectures (2D and 3D U-Net) for the segmentation of IVDs from MRI. The performance of these models was evaluated for morphological accuracy by comparing predicted IVD segmentations (Dice similarity coefficient, mDSC; average surface distance, ASD) to manual (ground truth) measures. Likewise, functional reliability and precision were assessed by evaluating the intraclass correlation coefficient (ICC) and standard error of measurement (SEm) of predicted and manually derived deformation measures. Results: Peak model performance was obtained using the 3D U-net architecture, yielding a maximum mDSC = 0.9824 and component-wise ASDx = 0.0683 mm; ASDy = 0.0335 mm; ASDz = 0.0329 mm. Functional model performance demonstrated excellent reliability ICC = 0.926 and precision SEm = 0.42%. Conclusions: This study demonstrated that a deep learning framework can precisely and reliably automate measures of IVD function, drastically improving the throughput of these time-intensive methods.http://www.sciencedirect.com/science/article/pii/S2665913123000456Medical image segmentationMachine learningComputer aided diagnosisLow back painIntervertebral disc degenerationIntervertebral disc mechanics |
| spellingShingle | James A. Coppock Nicole E. Zimmer Charles E. Spritzer Adam P. Goode Louis E. DeFrate Automated segmentation and prediction of intervertebral disc morphology and uniaxial deformations from MRI Osteoarthritis and Cartilage Open Medical image segmentation Machine learning Computer aided diagnosis Low back pain Intervertebral disc degeneration Intervertebral disc mechanics |
| title | Automated segmentation and prediction of intervertebral disc morphology and uniaxial deformations from MRI |
| title_full | Automated segmentation and prediction of intervertebral disc morphology and uniaxial deformations from MRI |
| title_fullStr | Automated segmentation and prediction of intervertebral disc morphology and uniaxial deformations from MRI |
| title_full_unstemmed | Automated segmentation and prediction of intervertebral disc morphology and uniaxial deformations from MRI |
| title_short | Automated segmentation and prediction of intervertebral disc morphology and uniaxial deformations from MRI |
| title_sort | automated segmentation and prediction of intervertebral disc morphology and uniaxial deformations from mri |
| topic | Medical image segmentation Machine learning Computer aided diagnosis Low back pain Intervertebral disc degeneration Intervertebral disc mechanics |
| url | http://www.sciencedirect.com/science/article/pii/S2665913123000456 |
| work_keys_str_mv | AT jamesacoppock automatedsegmentationandpredictionofintervertebraldiscmorphologyanduniaxialdeformationsfrommri AT nicoleezimmer automatedsegmentationandpredictionofintervertebraldiscmorphologyanduniaxialdeformationsfrommri AT charlesespritzer automatedsegmentationandpredictionofintervertebraldiscmorphologyanduniaxialdeformationsfrommri AT adampgoode automatedsegmentationandpredictionofintervertebraldiscmorphologyanduniaxialdeformationsfrommri AT louisedefrate automatedsegmentationandpredictionofintervertebraldiscmorphologyanduniaxialdeformationsfrommri |