Related Experiment Video
Updated: Mar 6, 2026
05:33
Three-Dimensional Shape Modeling and Analysis of Brain Structures
Published on: November 14, 2019
7.7K
Low-rank and sparse decomposition based shape model and probabilistic atlas for automatic pathological organ
Changfa Shi1, Yuanzhi Cheng2, Jinke Wang2
1Mobile E-business Collaborative Innovation Center of Hunan Province, Hunan University of Commerce, Changsha 410205, China; School of Computer Science and Technology, Harbin Institute of Technology, Harbin 150001, China.
Medical Image Analysis
|March 11, 2017
Summary
This study introduces an automatic method for accurate pathological organ segmentation in CT images using active shape models and low-rank decomposition. The approach robustly delineates organs, achieving accuracy comparable to human experts, even with severe pathology.
Area of Science:
- Medical Imaging
- Computer Vision
- Computational Anatomy
Background:
- Accurate human organ delineation is hindered by severe pathology and unclear organ borders in CT images.
- Existing segmentation methods struggle with corrupted data and complex anatomical variations.
Purpose of the Study:
- To develop an automatic, accurate, and robust method for pathological organ segmentation from CT images.
- To overcome limitations posed by severe pathology and improve organ boundary delineation.
Main Methods:
- Utilized the active shape model (ASM) framework combined with low-rank and sparse decomposition (LRSD) theory.
- Introduced a population-specific LRSD-based shape prior model (LRSD-SM) for handling gross errors and complex variations.
- Developed a patient-specific LRSD-based probabilistic atlas (LRSD-PA) for robust initialization and a hierarchical ASM search strategy for efficiency.
Main Results:
- The method ranked 3rd among state-of-the-art automatic methods in the SLIVER07 liver segmentation challenge.
- Demonstrated high accuracy and robustness in segmenting pathological livers and right lungs from 95 clinical CT scans.
- Achieved quantitative and qualitative results comparable to human expert delineation, even in highly severe cases.
Conclusions:
- The proposed method effectively addresses challenges in pathological organ segmentation.
- It offers a robust and accurate solution for delineating organ boundaries in the presence of significant pathology.
- The approach shows broad applicability for various pathological organs in clinical CT imaging.
