Automatic detection of calcified lesions in the descending aorta using contrast enhanced CT scans

In clinical practice, detection and quantification of calcified lesions in the coronary arteries and aorta is normally performed on non contrast enhanced Computed Tomography (CT) scans. This CT scan is generally followed by a contrast enhanced angiography (CTA) scan for better plaque visualization. We propose and evaluate a method for the detection of calcified lesions on CTA scans. Calcium lesion detection is straightforward on a CT scan, however performing the same task on a contrast enhanced angiography (CTA) scan is challenging, as the intensity of the contrast enhanced vessel lumen is very similar to that of calcium. Our method is based on semi-automatic centreline extraction and subsequent segmentation of high intensity objects to detect calcified lesions in the descending aorta. The method is evaluated on 19 datasets, and the results are compared to manual detection of calcifications on the CT scan. The proposed method has an accuracy of 82.4%, and 91% of the calcified lesions were successfully detected on the CTA scans.