Automated calcification detection and quantification in intravascular ultrasound images by adaptive thresholding

An innovative application of adaptive thresholding is used for calcification regions detection in intravascular ultrasound images. A priori knowledge about the acoustic shadow that usually follows the calcification regions is used as discriminant of other bright regions of the image. Tests were carried out with 20 in vivo coronary artery images obtained from different patients. This proposed algorithm presented specificity of 88% and sensitivity of 84%. A receiver operating characteristic (ROC) curve, whose area under the curve (AUC) was equal to 0.87, was plotted for evaluation of the algorithm performance. A technique for measure of the arc of calcium from the segmented images is also presented. The measures of the arc of calcium calculated by the proposed algorithm presented a correlation coefficient of 0.92 when compared with measures performed manually by a human expert.