Abstract
Abstract
Background
Minimally invasive thoracic surgery has become the standard of care, but is technically demanding, making objective assessment of technical competency essential during training. Video based
surgical gesture analysis
is a novel and objective method to evaluate technical skills during training for video-assisted thoracoscopic (VATS) lobectomies.
Methods
VATS lobectomy videos performed on a virtual reality simulator were annotated using a predefined
surgical gesture
taxonomy including blunt dissection, sharp dissection, ribbon use, and stapler use. Gesture metrics included duration, frequency, idle time, and entropy. Principal component analysis and a support vector machine were used to generate gesture-based performance scores. Validity evidence was assessed against simulator-derived metrics and human expert ratings using the VATS assessment tool (VATSAT).
Results
Surgical gesture analysis (SGA) was performed on 123 simulated VATS lobectomy procedures where 22 novices, 10 intermediates, and 9 experienced VATS surgeons performed three procedures each. Experienced surgeons had shorter procedure time (24.2 ± 9.2 min) than both intermediates (35.5 ± 9.9 min,
p
p
p
= 0.026) and novices (11 ± 3 s,
p
p
p
= 0.02) and experts (
p
= 0.049). Gestures correlated with simulator metrics and VATSAT scores, suggesting these reflected similar performance dimensions. SGA was able to classify experience level with a pass/fail threshold, achieving a sensitivity of 96.1% and specificity of 97.5%.
Conclusion
SGA can be used to assess technical performance for simulated VATS lobectomies. It was able to differentiate between experience levels for multiple parameters, and has potential advantages compared with existing assessments such as rating scales and simulator metrics. Further studies are needed to explore the applicability on real procedures and the clinical implications of Surgical Gesture Analysis.