Improved IVUS and VH-IVUS image segmentation using a hybrid approach based on active counter model and clustering algorithms

Heart attack due to the vulnerable atherosclerotic plaque is one of the most common causes of death in the world. One of the imaging modalities in the coronary artery is Virtual Histology Intravascular Ultrasound (VH-IVUS); however, segmentation of overlapped plaque components can be challenging. Therefore, this research study proposes a new approach based on the level set method to detect the plaque border in VH-IVUS images accurately. Three classifiers, including support vector machine (SVM), k-nearest neighbor, and proposed Selecting Nearest Pixel (SNP), were hybridized with FCM to enhance the VH-IVUS segmentation. Moreover, for segmentation and plaque extraction in IVUS images, a hybrid of the snake approach and fuzzy clustering method is proposed. Geometric features were extracted from VH-IVUS and IVUS images. The proposed hybrid model used 599 images obtained from 10 patients. The validation for the proposed segmentation model showed an average of 0.96 for the silhouette validity index. The SVM classifier is used to classify the TCFA and Non-TCFA plaques using VH-IVUS and IVUS features. The accuracy of the hybrid feature set was obtained over 0.99 for TCFA plaque.