Acquisition and usage of robotic surgical data for machine learning analysis.

Area of Science:

• Robotics in Medicine
• Surgical Training
• Artificial Intelligence in Healthcare

Background:

• Robot-assisted surgery (RAS) adoption necessitates efficient surgeon assessment methods.
• Current expert-based assessments are resource-intensive.
• Artificial intelligence (AI) offers a promising alternative, but lacks standardized data preparation protocols for clinical use.

Purpose of the Study:

• To develop and test a structured protocol for acquiring and preparing data for AI-based assessment of surgeons in robot-assisted surgery.
• To address the impediment to AI implementation in clinical settings due to a lack of standardized data preparation methods.

Main Methods:

• A structured guide was developed for data acquisition and preparation, including capturing surgical robot video data and surgeon 3D movement data.
• Data preparation involved steps such as capturing image data, extracting event data, capturing surgeon movement data, and annotating image data.
• The method was tested on porcine models using da Vinci Si and da Vinci Xi surgical robots.

Main Results:

• 188 videos (94 robot, 94 surgeon movement) were captured from 15 participants (11 novices, 4 experienced) performing 10 intra-abdominal RAS procedures.
• Event data, movement data, and labels were successfully extracted and prepared from the raw data for AI utilization.
• The study demonstrated the feasibility of collecting, preparing, and annotating diverse data types from robotic surgical systems.

Conclusions:

• The described methods enable the collection, preparation, and annotation of image, event, and motion data from surgical robotic systems.
• This structured approach facilitates the integration of AI for surgeon performance assessment in robot-assisted surgery.
• Standardized data preparation is crucial for advancing the clinical application of AI in surgical training and evaluation.