Vessel Wall Detection in the Images of Intravascular Optical Coherence Tomography based on the Graph Cut Segmentation

Intravascular Optical coherence tomography (IVOCT) is a catheter-based imaging method that employs near-infrared light to produce cross-sectional intravascular images. The size of lumen, where is known as vessel wall, is considered as the main indicator for artery clogging. The aim of this study is to investigate a new algorithm to extract lumen areas in IVOCT images which has time efficiency and high accuracy. The proposed algorithm consists of four main stages: pre-processing, feature extraction, kernel graph cut segmentation and finally the vessel wall detection and reconstruction. Generally, graph-based segmentation methods have a high precision in these images. So, we have used the kernel graph cut segmentation, which uses the piecewise model and Gaussian kernel function for the cost function. Furthermore, for initializing graph cut segmentation, we have used k-means clustering algorithm for the first frame. Since the amount of differences in the consecutive frames in IVOCT image collections is low, we have used the results of previous frame optimization as the initialization of the graph cut based segmentation of the current frame. The comparison results showed that suggested algorithm besides having a high precision (Hausdorff distance of 0.0653 mm and Mean distance of 0.0265 mm) also has a good processing speed in comparison with other methods (3.9 sec/frame).