Related Experiment Video
Updated: Nov 11, 2025
07:21
Magnetic Resonance Derived Myocardial Strain Assessment Using Feature Tracking
Published on: February 12, 2011
14.6K
Automated Artery Localization and Vessel Wall Segmentation using Tracklet Refinement and Polar Conversion
Li Chen1, Jie Sun2, Gador Canton2
1Department of Electrical and Computer Engineering, University of Washington, Seattle, WA, 98195, USA.
IEEE Access : Practical Innovations, Open Solutions
|March 29, 2021
Summary
This study introduces an automated system for artery segmentation, improving the analysis of blood vessel walls for cardiovascular risk assessment. The method enhances accuracy and reduces manual effort in segmenting atherosclerotic diseases.
Area of Science:
- Medical Imaging
- Cardiovascular Research
- Artificial Intelligence in Medicine
Background:
- Accurate quantitative analysis of blood vessel wall structures is crucial for studying atherosclerotic diseases and assessing cardiovascular event risks.
- Current computer-assisted segmentation tools often require manual preprocessing steps, limiting automation and efficiency.
- The inherent ring shape of vessel walls has not been fully leveraged in existing segmentation methods.
Purpose of the Study:
- To develop a fully automated system for artery localization and vessel wall segmentation.
- To overcome limitations of existing methods, including manual preprocessing and underutilization of vessel wall geometry.
- To improve the accuracy and robustness of vessel wall segmentation for clinical applications.
Main Methods:
- A novel system combining a neural network for artery centerline identification with a tracklet refinement algorithm for robust artery localization.
- Extraction of image patches from centerlines and conversion to a polar coordinate system for segmentation using 3D polar information.
- Development of a segmentation uncertainty score to identify potentially erroneous segmentations requiring manual review.
Main Results:
- The proposed system demonstrated superior automated vessel wall segmentation compared to traditional methods and standard convolutional neural network approaches on a large carotid artery dataset (>32000 images).
- The 3D polar coordinate-based segmentation effectively handled complex vessel geometries, contour discontinuities, and interference from neighboring vessels.
- The segmentation uncertainty score proved effective in identifying slices likely to contain errors, facilitating quality control.
Conclusions:
- The developed automated system offers a robust solution for artery localization and vessel wall segmentation, reducing the need for manual intervention.
- This technology has broad applications across different vascular beds, facilitating detailed vessel wall feature extraction and enhancing cardiovascular risk assessment.
- The system's ability to provide segmentation uncertainty aids in quality assurance and clinical adoption.
