Carotid Wall Segmentation and Imt Measurement in Longitudinal Ultrasound Images Using Morphological Approach

We developed a completely automated technique for carotid wall segmentation and intima – media thickness (IMT) measurement in ultrasound images based on the watershed transform. Morphological operators are used to extract the foreground markers for performing marker-based watershed segmentation. The watershed basin containing the carotid artery is automatically extracted. The far wall of the artery is segmented in order to measure IMT. Results obtained on 200 images were validated against human tracings. Performance was compared to a previously developed automated algorithm. Watershed technique outperformed the previous architecture and resulted in a IMT measurement error equal to 2.36 ± 2.76 pixels (0.15 ± 0.17 mm).