Study On Computer Aid Diagnosis Of Mass In Mammogram Base On Bilateral Comparison

With the social development, the incidence of breast disease showed a rising trend, this disease threats to women's health and even their lives seriously. The current study focus on early detection and prompts treatment of breast disease, early detection is important to reduce the mortality rate. At present, mammography is the main tool for imaging diagnosis of breast disease, and the breast Computer-Aided Diagnosis (CAD) system provides important diagnostic information to doctor, for improving the detection accuracy of breast disease.Breast mass is an important characterization of breast disease, but detection of mass is difficult in breast CAD system. In this paper, an automatic detection algorithm of breast mass was designed. It includes the initial detection algorithm of mass, texture features extraction, and Support Vector Machine (SVM) classification algorithm base on texture features.Firstly, according to symmetry of breast normal tissue and non-symmetry of mass, designed a mass initial detecting algorithm base on bilateral comparison, in this process, we got the initial detecting results of suspicious mass regions. Secondly, texture features of suspicious mass regions were extracted. According to the texture regularity between normal tissue and mass, calculated fractal dimension feature base on multi-level gray-scale map, and two-dimensional entropy feature. Experimental results showed, fractal dimension and two-dimensional entropy features can be showed the differences between normal tissue and mass well. Finally, used SVM to classify the kinds of suspicious mass regions, it is mass or not.In this paper, I randomly select the 106 X-ray mammograms to detect by my designed algorithm, achieved a positive fraction of 85.11% with average of 1.44 false positives per image. Experimental results demonstrate that the proposed detection algorithm is effective, it plays an important role to improve performance of CAD system.