AIxSuture: 314 Videos of Surgical Suturing Before and After Training

A collection of 314 five-minute videos showing 157 students performing surgical suturing before and after a one-hour training course. The dataset was created by Philipp Peters of RWTH Aachen University and includes expert assessments using the Objective Structured Assessment of Technical Skills (OSATS). Inter-rater variability for the assessments ranged from 0.8 to 0.83.

Use Cases

• Train computer vision models for automated skill assessment based on video recordings of suturing tasks.
• Benchmark AI-based assessment models against expert ratings using the OSATS global rating score.
• Analyze the impact of short training courses on suturing performance by comparing pre- and post-training videos.
• Study inter-rater reliability in surgical skill evaluation based on assessments from three experienced residents.

Strengths

• Contains 314 videos from 157 students, providing a paired pre- and post-training view.
• Expert annotations from three experienced oral and maxillofacial surgery residents using a standardized OSATS tool.
• High inter-rater reliability, with reported variability ranging from 0.8 to 0.83.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect a specific educational context and student population inherent to the source study.

Provenance

Source

RWTH Aachen University

Collection Method

Recorded from a randomized controlled trial on surgical training methods.