A Vascular Feature Detection and Matching Method Based on Dual-Branch Fusion and Structure Enhancement.

Area of Science:

• Medical Imaging
• Computer Vision
• Surgical Technology

Background:

• Laparoscopic 3D reconstruction faces challenges in obtaining internal cavity features and performing accurate image matching.
• Existing methods struggle with the precise detection and association of vascular features crucial for surgical navigation.

Purpose of the Study:

• To propose a robust method for detecting and associating vascular features for laparoscopic 3D reconstruction.
• To enhance the accuracy and repeatability of feature extraction and matching in minimally invasive surgery (MIS) images.

Main Methods:

• Developed a dual-branch weighted fusion vascular structure enhancement using Gaussian weighting, Frangi measure, and multiscale fractional anisotropic tensor (MFAT).
• Implemented a universal preprocessing framework for diverse MIS images.
• Utilized dual-circle detection with non-maximum suppression (NMS) for vascular feature extraction and zero sum of squared differences (ZSSD) for feature matching.

Main Results:

• The proposed method achieved an average accuracy score of 0.7149 and repeatability score of 0.5612 on the Vivo dataset.
• Demonstrated superior performance in quantity, repeatability, and accuracy of feature detection compared to existing methods.
• The dual-branch enhancement effectively addressed structural measurement differences between venous and microvessels.

Conclusions:

• The proposed vascular feature detection and association method significantly improves laparoscopic 3D reconstruction.
• The technique offers enhanced robustness and accuracy, providing a valuable tool for surgical navigation.
• The generalized preprocessing framework and advanced enhancement algorithm contribute to the method's effectiveness across various MIS scenarios.