Automatic detection of calcifications in the aorta from abdominal CT scans

An automatic method to detect calcifications in the aorta from CT scans of the abdomen is presented. Candidate objects are extracted by gray level thresholding. For each candidate object, a number of shape, spatial and gray level features are calculated. Based on those features, classification of candidate objects into calcifications and non-calcifications is performed in two stages. In the first stage, objects are discarded for which one of the features is not within a predefined range. In the second stage, classification of the remaining objects is performed using a k nearest-neighbor classifier. The method is evaluated on 20 scans containing different amounts of calcifications and gives high accuracy, sensitivity and specificity. In total, 119 calcifications out of 153 were detected at the expense of 33 false-positives.