Reducing the false positive rate of computer aided detection for CT colonography using Content Based Image Retrieval

We present a method to reduce the false positive rate of our computed tomographic colonography computer aided detection (CTC-CAD) system based on the paradigm of Content Based Image Retrieval (CBIR). Using the feature vectors generated from our CTC-CAD system in conjunction with the distance metric of our CBIR system we rank detections based on their likelihood of being true positive. By eliminating detections with low rank, we prune approximately half of the false-positive detections from our CAD system.