Local image registration a comparison for bilateral registration mammography

摘要: Early tumor detection is key in reducing the number of breast cancer death and screening mammography one most widely available reliable method for early detection. However, it difficult radiologist to process with same attention each case, due large amount images be read. Computer aided (CADe) systems improve rate; but current efficiency these not yet adequate correct interpretation CADe outputs requires expert human intervention. diagnosis (CADx) are being designed accuracy, they have been efficiently applied cancer. CADx can enhanced by considering natural mirror symmetry between right left breast. The objective this work evaluate co-registration algorithms accurate alignment enhancement. A set mammograms were artificially altered create a ground truth registration DEMONs , SPLINE deformable algorithms. accuracy was evaluated using mean square errors, mutual information correlation. results on 132 proved that over-perform DEMONS images.