A Heuristic Approach for Automated Mammogram Orientation in Computer Aided Diagnosis

Mammograms are the low dose X-ray images of the breast which are used to detect breast cancer. In Computer Aided Diagnosis (CAD) system for cancer detection, orientation of mammograms in a particular direction is important, so that for all the images, breast tissue should be displayed at similar position for the post-usage of CAD software. In proposed heuristic approach, pattern based radon transform is applied to detect the orientation of mammograms. The two orthogonal radon projections derive the orientation of chest wall and pectoral muscle. Depending upon input orientation to CAD, the proposed method flips or rotates mammogram. The algorithm was tested on 240 images, 100 images from miniMIAS, 100 computed radiography images and 40 full field digital mammogram images. For 98.75% of position invariant cases, orientation was detected and corrected. Due to its robustness, simplicity and fast speed, the method may be used for automatically detecting and correcting orientation of mammograms in CAD analysis.