Estimation of coronary atherosclerotic plaque composition based only on grey scale intravascular ultrasound images

In-vivo atherosclerotic plaque composition has been currently assessed by means of techniques based on spectral analysis of backscattered intravascular ultrasonic signals. Conversely, conventional IVUS images are based only on amplitude envelope of ultrasonic signals, discarding some frequency information and, consequently, disabling tissue characterization from spectral features as described previously. In this work, a computational tool has been developed for evaluation of coronary atherosclerotic plaque composition, without using of backscattered radiofrequency attributes. Textural analysis from atheromatous lesions has been combined with pattern recognition techniques in order to solve this problem. A preliminary test sample with 08 coronary arteries from 5 different patients, totalizing regions of interest, resulted into an average error-rate of 5.2%.