Multistage graph-based segmentation of thoracoscopic images.

Area of Science:

• Medical Imaging
• Computer Vision
• Surgical Technology

Background:

• Thoracoscopic surgery for scoliosis requires precise visualization of anatomical structures.
• Accurate segmentation of surgical video is crucial for real-time guidance and post-operative analysis.
• Existing methods may lack the precision needed for complex spinal procedures.

Purpose of the Study:

• To develop and evaluate a robust graph-based image segmentation method for thoracoscopic diskectomy procedures.
• To improve the accuracy and spatial coherence of segmented regions in surgical video.
• To provide a foundation for enhanced intraoperative guidance in scoliosis treatment.

Main Methods:

• Pre-processing of thoracoscopic images using Gaussian smoothing, contrast enhancement, and histogram thresholding.
• Application of a standard graph-based segmentation approach for initial coarse segmentation.
• Multistage region merging based on grey-level similarity, region size, and common edge length.

Main Results:

• The proposed method achieves good spatial coherence in segmented regions.
• Accurate localization of anatomical edges within the thoracoscopic images.
• Effective segmentation of regions of interest crucial for diskectomy procedures.

Conclusions:

• The multistage graph-based segmentation method offers a significant improvement for analyzing thoracoscopic surgical images.
• This technique enhances the precision of image segmentation, aiding in scoliosis treatment.
• The approach demonstrates potential for integration into advanced surgical navigation systems.