Despeckling and Segmentation of Ultrasound Images of Carotid Artery for Plaque Diagnosis in Health Care

A new method for segmentation of the carotid artery for classifying it as diseased or normal towards plaque diagnosis is proposed in this paper. Kernel Fuzzy CMeans clustering is used for segmenting the longitudinal section of the carotid artery using which the wall layers of the artery are identified for classifying it as diseased or normal. As a pre-processing step a–nonlinear mean filter is used for speckle filtering among a set of various traditional filters after measuring its performance. Removing speckle noise may remove useful information in the images like small lesions or artefacts, plaque or calcification in the arteries. Hence a filter that detects the speckle in each and every pixel and suppresses the speckled pixel is proposed in this paper. The algorithm was applied on ultrasound images of the carotid artery for extracting the boundary of the arterial wall and detecting the wall layers for clinical purposes. The results show excellent performance of the method.