Patient-Level Classification of Rotator Cuff Tears on Shoulder MRI Using an Explainable Vision Transformer Framework
Journal of Clinical Medicine · 23.01.2026 · 15 (3) · 928
DOI: 10.3390/jcm15030928 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Clinical Medicine, online erschienen am 23.01.2026. Verfasst von Murat Aşçı und 3 weiteren Autor:innen. Beteiligte Einrichtung: Department of Orthopedics and Traumatology, Faculty of Medicine, Bilecik Şeyh Edebali University, Bilecik 11230, Türkiye. Die Arbeit wurde bisher 5-mal zitiert (Crossref) und nennt 29 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Background/Objectives: Diagnosing Rotator Cuff Tears (RCTs) via Magnetic Resonance Imaging (MRI) is clinically challenging due to complex 3D anatomy and significant interobserver variability. Traditional slice-centric Convolutional Neural Networks (CNNs) often fail to capture the necessary volumetric context for accurate grading. This study aims to develop and validate the Patient-Aware Vision Transformer (Pa-ViT), an explainable deep-learning framework designed for the automated, patient-level classification of RCTs (Normal, Partial-Thickness, and Full-Thickness).
Methods
A large-scale retrospective dataset comprising 2447 T2-weighted coronal shoulder MRI examinations was utilized. The proposed Pa-ViT framework employs a Vision Transformer (ViT-Base) backbone within a Weakly-Supervised Multiple Instance Learning (MIL) paradigm to aggregate slice-level semantic features into a unified patient diagnosis. The model was trained using a weighted cross-entropy loss to address class imbalance and was benchmarked against widely used CNN architectures and traditional machine-learning classifiers.
Results
The Pa-ViT model achieved a high overall accuracy of 91% and a macro-averaged F1-score of 0.91, significantly outperforming the standard VGG-16 baseline (87%). Notably, the model demonstrated superior discriminative power for the challenging Partial-Thickness Tear class (ROC AUC: 0.903). Furthermore, Attention Rollout visualizations confirmed the model’s reliance on genuine anatomical features, such as the supraspinatus footprint, rather than artifacts.
Conclusions
By effectively modeling long-range dependencies, the Pa-ViT framework provides a robust alternative to traditional CNNs. It offers a clinically viable, explainable decision support tool that enhances diagnostic sensitivity, particularly for subtle partial-thickness tears.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Orthopedics and Traumatology, Faculty of Medicine, Bilecik Şeyh Edebali University, Bilecik 11230, Türkiye.
- Department of Software Engineering, Faculty of Engineering and Architecture, Eskişehir Osmangazi University, Eskişehir 26040, Türkiye.
- Center of Intelligent Systems Applications Research, Eskişehir Osmangazi University, Eskişehir 26040, Türkiye.
- Department of Computer Engineering, Faculty of Engineering and Architecture, Eskişehir Osmangazi University, Eskişehir 26040, Türkiye.
- Department of Radiology, Yunus Emre State Hospital, Ministry of Health, Eskişehir 26190, Türkiye.
ORCID-Profile: Murat Aşçı, Sergen Aşık, Ahmet Yazıcı, İrfan Okumuşer
Zitieren
APA
Aşçı, M., Aşık, S., Yazıcı, A., & Okumuşer, İ. (2026). Patient-Level Classification of Rotator Cuff Tears on Shoulder MRI Using an Explainable Vision Transformer Framework. Journal of Clinical Medicine, 15(3), 928. https://doi.org/10.3390/jcm15030928
Vancouver
Aşçı M, Aşık S, Yazıcı A, Okumuşer İ. Patient-Level Classification of Rotator Cuff Tears on Shoulder MRI Using an Explainable Vision Transformer Framework. Journal of Clinical Medicine. 2026;15(3):928. doi:10.3390/jcm15030928
BibTeX
@article{A2026,
author = {Aşçı, Murat and Aşık, Sergen and Yazıcı, Ahmet and Okumuşer, İrfan},
title = {Patient-Level Classification of Rotator Cuff Tears on Shoulder MRI Using an Explainable Vision Transformer Framework},
journal = {Journal of Clinical Medicine},
year = {2026},
volume = {15},
number = {3},
pages = {928},
doi = {10.3390/jcm15030928}
}
EUVIMED