System for Central Venous Catheterization Training Using Computer Vision-Based Workflow Feedback.

Area of Science:

• Medical simulation and training
• Artificial intelligence in healthcare
• Medical procedure automation

Background:

• Central venous catheterization is a critical medical procedure requiring skilled performance.
• Traditional training methods often rely on expert supervision, limiting accessibility and scalability.
• There is a need for automated systems to provide consistent, real-time feedback during training.

Purpose of the Study:

• To develop and evaluate an automated training system for central venous catheterization.
• To provide real-time instruction and feedback to trainees without requiring an expert observer.
• To leverage video-based workflow recognition and electromagnetic tracking for enhanced training.

Main Methods:

• Utilized a combination of convolutional neural networks (CNN) and recurrent neural networks (RNN) for video-based workflow recognition.
• Integrated electromagnetic tracking for precise procedural assessment.
• Evaluated system performance through task recognition accuracy, transitional delay, and participant usability questionnaires.

Main Results:

• The system achieved an 86.2% accuracy in recognizing tasks within the central venous catheterization workflow.
• An average signed transitional delay of -0.7 seconds was recorded, indicating near real-time feedback.
• Participants reported high usability, with an overall score of 4.7 out of 5, and found the interactive task list particularly valuable (4.8/5).

Conclusions:

• The developed system effectively provides meaningful instruction and feedback for central venous catheterization training.
• The system enhances training accessibility by removing the need for constant expert observation.
• Participants perceived the system as a valuable tool for improving central venous catheterization skills.