Automated segmentation of media-adventitia and lumen from intravascular ultrasound images using non-parametric thresholding

This paper presents an automatic method for detecting lumen and media-adventitia boundaries from "Intravascular Ultrasound Image" (IVUS). The linear stretching equation is proposed to enhance the contrast of IVUS image. A complexity reduction technique for IVUS image is conducted by an adaptive k-mean clustering. Finally, conventional integrated processes based on mathematics morphology and convex hull technique are applied to estimate boundaries of lumen and media-adventitia. To evaluate the performance of the proposed method, 20 IVUS images from Simone Bolocco's data set was considered. The experiment provides satisfied results with high accuracy. Moreover, the experimental results show our approach outperform empirical thresholding method.