Vision transformer-based ordinal model for semi-quantitative synovitis grading on musculoskeletal ultrasound: a retrospective study with internal and external validation.

A Vision Transformer-based model achieved accurate and clinically applicable synovitis grading on musculoskeletal ultrasound, outperforming traditional CNNs in a retrospective study with internal and external validation.

The study included 312 patients, 624 joints, and 1,248 ultrasound images, which were divided into training, validation, internal test, and external test cohorts.

A Swin Transformer-based ordinal grading model was trained on gray-scale and power Doppler ultrasound data and compared with CNN baselines using internal validation and external testing.

The Vision Transformer achieved the best overall performance, with an internal-test accuracy of 0.809 and AUC of 0.929, while evaluation on an external test set remained acceptable with an accuracy of 0.771 and AUC of 0.904.