A 3D Deep Learning Approach for Meniscus Tear Severity at the Region-level
Berrimi, Mohamed; Oussalah, Mourad; Jennane, Rachid (2023-11-21)
IEEE
21.11.2023
M. Berrimi, M. Oussalah and R. Jennane, "A 3D Deep Learning Approach for Meniscus Tear Severity at the Region-level," 2023 Twelfth International Conference on Image Processing Theory, Tools and Applications (IPTA), Paris, France, 2023, pp. 1-6, doi: 10.1109/IPTA59101.2023.10320051.
https://rightsstatements.org/vocab/InC/1.0/
© 2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
https://rightsstatements.org/vocab/InC/1.0/
© 2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
https://rightsstatements.org/vocab/InC/1.0/
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:oulu-202501031043
https://urn.fi/URN:NBN:fi:oulu-202501031043
Tiivistelmä
Abstract
Meniscus tears are a common knee joint injury and a significant risk factor for the development of knee OsteoArthritis (OA). Accurate detection of meniscus tears and classification of their severity is crucial for early intervention and treatment planning. In this study, we propose to use 3D deep learning models for the detection and classification of the severity of meniscus tears in both the lateral and medial regions of the knee on 3D MRI scans. Several models were trained and evaluated on a dataset comprising 2400 MRI scans from the Osteoarthritis Initiative (OAI) database. The dataset includes diverse range of meniscus tear cases, which provides a representative sample for robust model training and evaluation. The 3D deep learning models leverage the spatial information present in the MRI images to capture complex features revealing meniscus tears. For the detection of the tear region, our proposed model achieves a promising AUC score of 91.9%. The models successfully identified the presence of meniscus tears in both lateral and medial regions, demonstrating their effectiveness in localizing the tear within the knee joint. Furthermore, in terms of classification of the identified tear region, our model achieved an AUC score of 79.8%, exhibiting relatively high performance rate and demonstrating the potential of the proposed approach as a valuable tool for meniscus tear detection and classification.
Meniscus tears are a common knee joint injury and a significant risk factor for the development of knee OsteoArthritis (OA). Accurate detection of meniscus tears and classification of their severity is crucial for early intervention and treatment planning. In this study, we propose to use 3D deep learning models for the detection and classification of the severity of meniscus tears in both the lateral and medial regions of the knee on 3D MRI scans. Several models were trained and evaluated on a dataset comprising 2400 MRI scans from the Osteoarthritis Initiative (OAI) database. The dataset includes diverse range of meniscus tear cases, which provides a representative sample for robust model training and evaluation. The 3D deep learning models leverage the spatial information present in the MRI images to capture complex features revealing meniscus tears. For the detection of the tear region, our proposed model achieves a promising AUC score of 91.9%. The models successfully identified the presence of meniscus tears in both lateral and medial regions, demonstrating their effectiveness in localizing the tear within the knee joint. Furthermore, in terms of classification of the identified tear region, our model achieved an AUC score of 79.8%, exhibiting relatively high performance rate and demonstrating the potential of the proposed approach as a valuable tool for meniscus tear detection and classification.
Kokoelmat
- Avoin saatavuus [38865]