Automatic detection and classification of malignant tumor in mammograms image using image feature fractal dimension

Breast cancer is a second largest disease in worldwide and even in India as per the statistics of world health organization after the lung cancer. The conventional approach to identify the breast cancer is biopsy, it takes on an average of more than week together time and most of the hospitals do not have this facility to perform the biopsy. This approach also demands expertise in the domain of analysis of tumor tissues to identify the cancerous cell. Hence to overcome the drawbacks of the conventional diagnosis system. In this paper a novel approach has been presented to diagnose the breast cancer by analyzing X-ray mammograms by a technique called rotational contour based fractal dimension with an interval of 60 degree. In this paper, the work is categorized into four phases,(1).Enhancing the mammogram images using Gabor filter and also estimated PSNR before and after the enhancement of the mammogram images that leads to accurate segmentation of tumor from the mammogram. (2).The automatic segmentation of region of the tumor through watershed and morphological operations and also obtained the contour of the tumor. (3). The contour analysis has been performed using a new approach called contour based fractal dimension approach that gives excellent classification result for the benign and malignant tumor. The Fractal Dimension for benign tumor ranges from 1.462 to 1.71 where as for malignant tumor the FD ranges from 1.78 to 3.78. And the Standard Deviation for benign tumor 0.06 and for malignant is 0.58. (4).In classification phase the automatically identifying and segregating the cancer disease. In this work, consider huge set of images from publicly available popular databases such as Digital Datagram Screening Mammogram, MIAS and also considered for the images available in the SDM hospital, Ujire and Dharwad etc. This approach gives almost 100 percent accuracy. Hence this technique can be considered as diagnostic parameter for the identification and classification of disease which serves the oncologist to take better decision.