Extra-abdominal trocar and instrument detection for enhanced surgical workflow understanding

Area of Science:

• Minimally Invasive Surgery
• Surgical Robotics
• Computer-Aided Surgery

Background:

• Intra-abdominal instrument tracking is limited to visible instruments.
• Understanding instrument location and status is crucial for surgical workflow.
• Existing solutions often lack comprehensive tracking capabilities.

Purpose of the Study:

• To develop an extra-abdominal camera system for tracking surgical instruments.
• To detect trocars and determine their occupancy state during laparoscopic surgery.
• To enhance laparoscopic workflow analysis by providing additional instrument location data.

Main Methods:

• A dataset of four laparoscopic surgeries was recorded using synchronized extra-abdominal 2D cameras.
• A deep learning network was trained for trocar detection and occupancy state classification.
• The network included a trocar detection module, centroid tracker, and temporal model.

Main Results:

• The trocar detection model achieved an F1 score of [insert score].
• The occupancy state prediction model achieved an F1 score of [insert score].
• These results represent a significant step towards improved surgical workflow understanding.

Conclusions:

• The proposed extra-abdominal tracking method shows promise for monitoring trocars and their occupancy.
• Future work involves expanding the dataset and integrating intra-abdominal imaging.
• Accurate instrument-to-trocar assignment can be facilitated by combining both tracking approaches.