Extracting Significant Features from Virtual Histology to Detect Vulnerable Plaque

One of the major challenges and concerns of researchers is an early detection and diagnosis of thin-cap-fibro-atheroma or vulnerable plaque to prevent the sudden heart events. Recently, Virtual Histology (VH) as a new approach based on spectral analysis of Intravascular Ultrasound (IVUS) provides color coded of coronary tissue maps. In IVUS-VH image, plaque's components can be discriminated based on echogenicity. Nonetheless; available methods do not provide clinical relevant information about the pattern of plaque structure, plaque composition, and geometric position of each components, location or distribution of plaque components toward the lumen border. In this paper, we proposed a new method of feature extraction that plays a decisive role in vulnerable plaque detection including NCCL (Necrotic Core in Contact with the Lumen), DCCL (Dense Calcium in Contact with the Lumen), Confluent NC and Confluent DC.