"Slow is Smooth, Smooth is Fast”: AI Analysis of Suturing at Depth Utilizing a Novel Simulator

Abstract

Objective: To develop and validate a novel suturing training simulator for cardiothoracic surgeons in training, using artificial intelligence (AI)-assisted computer vision for objective assessment of technical performance across different levels of surgical expertise. Methods: A 3D-printed dome-shaped suturing simulator was created with circular silicone targets positioned at angles to replicate five distinct anatomical planes relevant to cardiothoracic surgery. Twenty participants (10 attending cardiac surgeons and 10 surgical trainees) were recorded while performing running baseball sutures on two distinct planes. Motion tracking using MediaPipe, an open-source computer vision framework, quantified temporal and spatial parameters of hand movement, including completion times, aggregate hand path lengths, movement smoothness and efficiency. Computer vision image analysis was used to evaluate suture uniformity and precision, with uniformity measured using the standard deviation (SD) of distances between consecutive sutures. Group differences were analyzed using Student's t-test with Bonferroni correction for multiple comparisons. Results: AI-assisted computer vision analysis differentiated experts from trainees with high precision. Attending cardiac surgeons performed significantly faster than trainees in needle loading (1.9s vs. 2.5s, p 0.01) and stitch execution (4.6s vs. 6.4s, p 0.01), resulting in shorter ring completion times (252s vs. 312s, p 0.02). Attending surgeons also demonstrated superior motion efficiency and less irregular movements during key maneuvers, characterized by reduced overall path length (6,714 vs. 10,154 mm, Bonferroni corrected p = 0.03), slower accelerations (20 vs. 25 mm/s2, Bonferroni corrected p = 0.03) and reduced vibrational motion patterns (26,786,414 vs. 56,008,703 mm2/s4, Bonferroni corrected p = 0.03). Image analysis confirmed that attending surgeons produced significantly more uniform sutures with lower variability in inter-stitch distances (SD 1.72 vs. 2.14 mm, p = 0.02; 67.8% vs. 59.4%, p = 0.01). Conclusions: AI-assisted suturing assessment reliably distinguished expert from novice cardiothoracic surgeons using quantifiable measures. Experts performed more deliberate, controlled and smoother movements, reinforcing the surgical teaching that "slow is smooth, and smooth is fast." The integration of computer vision analysis with a cost-effective, anatomically relevant simulator represents a scalable platform for objective skill assessment and personalized feedback in surgical training.