Carotid wall volume quantification from magnetic resonance images using deformable model fitting and learning-based...

Area of Science:

• Medical Imaging
• Cardiovascular Research
• Biomedical Engineering

Background:

• Accurate quantification of carotid vessel wall volume is crucial for assessing cardiovascular disease risk.
• Manual segmentation methods are time-consuming and prone to inter-observer variability.
• Developing automated techniques can improve efficiency and consistency in clinical practice.

Purpose of the Study:

• To present and validate an automated method for carotid vessel wall volume quantification using magnetic resonance imaging (MRI).
• To compare the performance of the automated method against manual measurements.
• To make image data, annotations, and measurements publicly available for research.

Main Methods:

• A novel method combining lumen and outer wall segmentation using deformable models and a learning-based correction step.
• Automatic vessel wall volume determination in the carotid bifurcation region after manual initialization.
• Training and evaluation on datasets from a population-based elderly study with manual annotations by observers.

Main Results:

• The automated method demonstrated performance comparable to manual measurements for wall volume and normalized wall index.
• The technique successfully quantified carotid vessel wall volume in a large dataset.
• Reproducibility and accuracy were validated against expert manual segmentations.

Conclusions:

• The developed automated method provides a reliable and efficient approach for carotid vessel wall volume quantification.
• This technique has the potential to enhance the assessment of atherosclerosis and cardiovascular risk.
• Public data sharing facilitates further research and development in the field.