Semi-automatic Registration of Mammographic Images for Breast Cancer Progression Assessment

Breast cancer is the most common cancer among women. Despite significant advances in the medical field and its related technologies, cancer remains one of the biggest problems in medical science. Point common in all cancers need to do surgery and remove the tumor in advanced cases. Unfortunately, the surgery is not successful in some cases and later re-released in the body. Sometimes changing in the patient''s tissues hides from human sight. One of the problems of comparing images taken at different times is change in imaging condition that causes distortion in the image and the difference between the textures in the image. To solve this problem and reduce the error, in this paper a semi-automatic method is presented for mammographic images registration. In this method using feature extraction of a mammogram such as brightness of the image, transformation distance, and also information that the user presents, can obtain the actual relationship between images and determines the differences between the images with less error. The proposed method will run on a sample of 27 mammograms from the Parto Teb Azma Imaging Center. We will show that in addition to maximizing similarities and minimizing differences between two images, the second image retains its overall shape. Generally correlation coefficient of mammograms after image registration in the semi-automatic method increase to 0.2670 and the difference squared in the semi-automatic method will decrease to 0.0968.