Automatic soft and hard plaque detection in IVUS images: A textural approach

This paper describes an automated algorithm for plague detection in Intra Vascular Ultrasound images, using an adaptive border detection method for selection Region Of Interest (RIO) and detect the soft and hard plague. As shadow appears behind the calcification plaque, it makes it difficult especially for soft plague. Our algorithm divided by two mail part; one, border detection and selection the ROI and second, plague detection. Therefore we propose to GLCM and FCM for part one and FCM for part two. Meanwhile we use carve fitting, morphology, noise reducing, averaging and some other function to achieve a higher level of accuracy. Results show that the proposed meth efficiently detected ROI and then hard plague also soft plague even in complicated images. The proposed algorithm presented specificity of 83% and a sensitivity of 91% as a result of 60 different IVUS images from seven different patient cases.