An approach for detection of edges in carotid ultrasound images and analysis of intima-media complex using morphological features

Atherosclerosis is a complex vascular disease that causes the condition of narrowing, stiffening and hardening of artery walls and it's the first clinical manifestation of cardiovascular disease that leads to heart attack and stroke. Progression of atherosclerosis is marked by intima-media thickness which is proven to be an early indicator of the disease. In this work, an attempt is made to preserve the edges of intima-media complex using nonlinear total variation diffusion filter. The performance of segmentation method and the results are validated against ground truth values. Geometric features are extracted from the segmented region and statistical analysis is performed. Bland-Altman plots show that, the values between 95% confidence interval are with overall good fit with minimal bias between segmentation method and ground truth (manually segmented by an expert). The analysis of features plays clinically a significant role in finding the pathological conditions of carotid arteries.