VasSurDetect: An AI-based Chatbot for Rapid, Photo-based Feedback in Microsurgical Suturing Training

Area of Science:

• Medical Simulation
• Artificial Intelligence in Medicine
• Surgical Training Technology

Background:

• Microsurgical suturing requires consistent, timely feedback for skill development.
• Current faculty feedback is often delayed, subjective, and limited by instructor availability.
• An AI-powered mobile application was developed to address these limitations.

Purpose of the Study:

• To develop and evaluate an AI chatbot, VasSurDetect, for automated, rapid feedback on microsurgical suturing quality.
• To provide objective, quantitative metrics on suturing technique using image analysis.
• To enhance deliberate practice in simulation-based microsurgical training.

Main Methods:

• Collected 350 paired microscope images of simulated vascular sutures (chicken wing model).
• Annotated images for knots, purchase segments, and suture material.
• Trained and compared nine object detection models (e.g., YOLOv10, YOLOv8) to quantify suturing parameters.
• Deployed the selected model into a chatbot workflow for automated analysis and reporting.

Main Results:

• YOLOv10 achieved the highest F1 score (approx. 0.89) for inside-view suture detection.
• The system generated stitch-level measurements (spacing, angle, leftover length) with high accuracy (e.g., spacing error ~0.4 mm).
• Automated feedback reports were delivered within 10 seconds.

Conclusions:

• VasSurDetect offers rapid, standardized, image-based feedback for microsurgical suturing practice.
• The AI tool requires no additional hardware and complements expert supervision.
• It supports high-frequency deliberate practice and objective performance tracking in training.