A Deformable Model-based Minimal Path Segmentation Method for Kidney MR Images.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computational Anatomy

Background:

• Accurate segmentation of organs in medical imaging is crucial for quantitative analysis.
• Existing segmentation methods may struggle with complex anatomical structures or image artifacts.
• Automated segmentation reduces inter-observer variability and processing time.

Purpose of the Study:

• To develop and validate a novel minimal path segmentation method for mouse kidney MR images.
• To automate the segmentation process, improving efficiency and accuracy.
• To assess the method's robustness and generalizability to other organs.

Main Methods:

• Utilized dynamic programming and a minimal path segmentation approach on weighted graphs.
• Developed an energy function combining distance and gradient information to guide segmentation.
• Implemented an algorithm for automatic initial endpoint placement and optimization using dynamic programming.
• Employed Principle Component Analysis (PCA) to create a deformable model for prior knowledge integration.

Main Results:

• The method achieved a mean overlap ratio of 95.19% ± 0.03% compared to manual segmentation.
• The mean distance error between automatic and manual segmentation was 0.82 ± 0.41 pixels.
• The segmentation process was demonstrated to be fast, accurate, and robust across 44 mouse kidney MR images.

Conclusions:

• The developed minimal path segmentation method provides a fast, accurate, and robust solution for mouse kidney MR image analysis.
• The approach demonstrates high concordance with manual segmentation, suggesting clinical relevance.
• The method's adaptability indicates potential for application in segmenting other organs in medical imaging.