Evaluation framework for carotid bifurcation lumen segmentation and stenosis grading
K Hameeteman1, M A Zuluaga, M Freiman
1Biomedical Imaging Group Rotterdam, Departments of Radiology & Medical Informatics, Erasmus MC, P.O. Box 2040, 3000 CA Rotterdam, The Netherlands. K.Hameeteman@Erasmusmc.nl
Medical Image Analysis
|March 23, 2011
Summary
A new framework enables objective comparison of carotid artery segmentation algorithms. Automated methods achieve precision comparable to manual analysis, advancing clinical assessment.
Area of Science:
- Medical Imaging
- Computer-Aided Diagnosis
- Biomedical Engineering
Background:
- Accurate carotid artery lumen segmentation and stenosis grading are crucial for cardiovascular disease assessment.
- Existing methods lack standardized, objective quantitative comparison, hindering progress.
Purpose of the Study:
- To introduce a standardized evaluation framework for carotid artery lumen segmentation and stenosis grading algorithms.
- To quantitatively compare the performance of automated algorithms against a reference standard.
Main Methods:
- Development of a data repository with 56 multi-center, multi-vendor CTA datasets.
- Establishment of a reference standard for lumen segmentation and stenosis grading.
- Definition of standardized evaluation measures for algorithm comparison.
Main Results:
- The proposed framework facilitated a standardized, objective quantitative comparison of eight participating teams' algorithms.
- Automated carotid artery lumen segmentation demonstrated precision comparable to manual annotation.
- The evaluation framework was successfully introduced at the MICCAI 2009 workshop.
Conclusions:
- The developed framework provides a robust platform for evaluating carotid artery segmentation and stenosis grading algorithms.
- Automated segmentation shows potential for clinical application, matching manual annotation accuracy.
- The framework remains open for further submissions to drive innovation.
