Video content analysis of surgical procedures

Area of Science:

• Computer Vision
• Medical Informatics
• Surgical Technology

Background:

• Minimally invasive surgery generates valuable video data for applications beyond immediate patient care.
• Surgical videos can be archived for cognitive training, skills assessment, and workflow analysis.
• Video content analysis methods are increasingly adapted for the surgical domain.

Purpose of the Study:

• To review recent developments in content-based video analysis of surgical operations.
• To analyze future research directions in this rapidly expanding field.

Main Methods:

• A comprehensive literature search was conducted using PubMed and Google Scholar.
• Keywords included 'surgery', 'video', 'phase', 'task', 'skills', 'event', 'shot', 'analysis', 'retrieval', 'detection', 'classification', and 'recognition'.
• Articles were categorized by technical goal, surgery type, and operational structure.

Main Results:

• 81 articles were included, with over 50% published in the last three years, indicating rapid growth.
• Significant research exists for video task detection and retrieval in eye surgery.
• Endoscopic surgery research is diverse, covering gesture/task classification, skills assessment, tool recognition, and event detection, with deep neural networks showing promise.

Conclusions:

• Content-based video analysis of surgical operations is a dynamic and expanding research area.
• Future research should focus on areas like shot boundary detection, keyframe extraction, summarization, pattern discovery, and annotation.
• Development of standardized benchmark datasets is crucial for algorithm evaluation and comparison.