Carotid stenosis assessment with multi-detector CT angiography: comparison between manual and automatic segmentation
Chengcheng Zhu1, Andrew J Patterson, Owen M Thomas
1Department of Radiology, Cambridge University Hospitals Foundation Trust, University of Cambridge, Box 218, Cambridge CB2 0QQ, UK.
The International Journal of Cardiovascular Imaging
|November 9, 2012
Summary
Automated segmentation of carotid artery disease is more reproducible than manual methods for quantifying luminal stenosis. This improves accuracy in selecting optimal management strategies for patients.
Area of Science:
- Medical Imaging
- Radiology
- Cardiovascular Disease
Background:
- Carotid artery disease management relies on accurate luminal stenosis quantification.
- Multi-Detector CT angiography (MDCTA) is a key imaging modality.
- Reproducibility of stenosis measurements is crucial for clinical decision-making.
Purpose of the Study:
- To evaluate the reproducibility of carotid stenosis quantification.
- To compare manual versus automated segmentation methods.
- To assess measurement consistency using submillimeter resolution MDCTA.
Main Methods:
- 35 patients with carotid artery disease underwent contrast-enhanced MDCTA.
- Two readers manually quantified stenosis from axial images and MIP reconstructions.
- Automated segmentation was performed using the Vascular Modelling Toolkit (VMTK).
- NASCET criteria were applied for stenosis measurement.
Main Results:
- Good agreement was observed across manual and automated methods.
- Automated segmentation demonstrated superior inter-observer agreement (ICC=0.99) compared to manual axial (ICC=0.82) and MIP (ICC=0.86) measurements.
- Higher reproducibility was achieved with automated carotid stenosis quantification.
Conclusions:
- Automated segmentation offers higher reproducibility for carotid stenosis quantification than manual methods.
- This finding supports the clinical utility of automated tools for managing carotid artery disease.
- Improved reproducibility can enhance patient management strategies.
