Automatic Vessel Detection Technique in Laparoscopic Surgical Imaging Including Surgical Instruments

Area of Science:

• Medical Imaging
• Robotic Surgery
• Computer Vision

Background:

• Accurate identification of blood vessels is crucial in minimally invasive surgery.
• Existing methods may require specialized hardware or complex parameter tuning.

Purpose of the Study:

• To develop a novel, hardware-independent technique for blood vessel recognition in robot-assisted laparoscopic surgery.
• To improve surgical safety and precision by accurately segmenting vascular structures.

Main Methods:

• Utilized Hessian matrix in HSV color space for blood vessel edge detection.
• Employed histogram equalization, clustering, and region growing to differentiate vessels from surgical instruments.
• Processed real robot-assisted laparoscopic surgery video data.

Main Results:

• Achieved an average recall of 92.67% for vessel detection.
• Demonstrated efficient processing speeds: ~0.3s/frame at 640p and ~0.8s/frame at 1280p.
• The method requires no additional hardware or parameter adjustments.

Conclusions:

• The proposed color-based technique effectively recognizes blood vessels in surgical videos.
• This approach offers a practical and efficient solution for intraoperative vascular visualization.
• The method enhances surgical instrument removal and vessel segmentation accuracy.