Research On Image Enhancement Algorithm For Mammography

Computer-aided detection and diagnosis system of breast cancer is an important auxiliary tool to early breast cancer diagnosis. The mammography enhancement technique is an important research field of computer-aided detection and diagnosis system, whose purpose is to enhance the detection results of breast cancer.To solve the problem of fuzzy, noise and low resolution of mammography, a window wide and window level technology, a pseudo color image processing technology, image amplifier technology and a gray level transportation technology are developed to enhance the x-ray mammography. By using these image enhancement tools, the users can observe the region-of-interest (ROI) and the gray-scale of the mammography more clearly and efficiently to find the cancer area of mammography.To solve the problem of higher false-positive rates, this thesis designs a new technology which groups the enhancement technology of piecewise-linear and unsharp transportation. After using the multi-level threshold selection to get the breast region, the new technology is used to enhance the mammography, and then the ROI containing suspicious mass is extracted by using a multi-scale template matching procedure. When the sensitivity is identical, the result shows that using image enhancement preprocessing is better than extracting ROI from the original mammography directly and the false-positive is reduced a lot.To raise the accuracy of segmentation,image enhancement preprocessing which include close neighbor tissue restrained, background region corrected, gamma and Gauss filter is used to deal with ROI. Then a dynamic programming-based method is proposed to suspicious mass segmentation. The experiment compares the overlay ratio between the mass segmentation and the doctor circled. The result shows that using preprocessing enhancement technology gets better segmentation performance.