Research on workflow recognition for liver rupture repair surgery

Area of Science:

• Medical technology
• Artificial intelligence in surgery
• Surgical workflow analysis

Background:

• Liver rupture can lead to critical conditions like hemorrhage and shock.
• Surgical workflow recognition in liver rupture repair surgery is crucial for reducing errors and improving patient outcomes.
• Accurate identification of surgical phases aids in training and real-time performance monitoring.

Purpose of the Study:

• To develop an automated system for surgical workflow recognition in liver rupture repair surgery.
• To introduce and evaluate a novel self-attention-based recurrent convolutional neural network (SA-RLNet) for this task.
• To assess the accuracy and generalization capabilities of the proposed model on a dedicated surgical dataset.

Main Methods:

• A dataset of 45 liver rupture repair surgery videos was created, featuring nine surgeons.
• An end-to-end deep learning model, SA-RLNet, incorporating a self-attention mechanism, was developed.
• The model was trained and validated for surgical phase classification.

Main Results:

• The SA-RLNet achieved a high accuracy of 90.6% in surgical phase classification.
• The self-attention mechanism effectively identified important features and their relationships.
• The model demonstrated strong generalization capabilities across the dataset, capturing subtle phase variations.

Conclusions:

• The SA-RLNet approach shows significant promise for automated surgical workflow recognition in liver rupture repair.
• The developed system can accurately classify surgical phases, contributing to enhanced surgical quality and safety.
• This technology has feasible applications in real-world surgical settings, aiding surgeons and improving patient care.