Computer-Based Tracking of Microsurgical Instruments: A Novel Assessment Tool for Robot-Assisted and Conventional...

Area of Science:

• Microsurgical skills assessment
• Surgical robotics
• Computer-assisted surgery

Background:

• Objective self-assessment tools are crucial for microsurgical training and resource management.
• Clinical integration of robotic microsurgery necessitates validated training and evaluation methods.
• Current assessment methods lack efficiency and objectivity for both conventional and robotic procedures.

Purpose of the Study:

• To develop and validate computer algorithms for objective self-assessment of microsurgical skills.
• To enable direct comparison between conventional and robotic-assisted microsurgical techniques.
• To enhance the efficiency and accessibility of microsurgical training.

Main Methods:

• Development of two deep convolutional neural network-based algorithms for instrument tracking.
• Application of supervised and semi-supervised learning on 84 microsurgical training videos.
• Statistical analysis including t-test, ANOVA, linear regression, and correlation.

Main Results:

• Algorithms successfully track conventional and robotic microinstruments in training videos.
• Total trajectory length correlated with procedure time and skill assessment, indicating operative efficiency.
• Robotic-assisted procedures showed longer duration for experienced surgeons but reduced hand tremor across all groups.

Conclusions:

• Developed algorithms provide objective, efficient, and accessible self-assessment in microsurgery.
• The tools allow for the first direct comparison of robotic-assisted versus conventional microsurgical performance.
• This advancement supports improved training and evaluation in microsurgery.