Research Of Microcalcification Cluster Detection Technology In Mammograms

Breast cancer is a common form of malignant tumors which threaten women health, there is a increasing trend in its incidence recent years. It shows clearly that early discovery, early diagnosis and early treatment of breast cancer can significantly increase the chance of survival and reduce the mortality of breast cancer for patients. Mammography has become the most common mean because of its convenience and efficiency in the current breast cancer screening. The microcalcifications which are small and granular in mammograms are an important early sign for early breast cancer. It is estimated that about 30% to 50% of breast cancers detected radiographically demonstrate microcalcifications in mammograms. One of the key technology is to find microcalcifications in mammograms in time and judge whether they will become malignant or not in early diagnosis.Because the gray values are close between the microcalcifications and the breast tissue in mammograms, viewed mammograms display only a very small part of the total information they contain, most of the information should not be detected by experts and doctors. We can detect the microcalcifications by using digital image processing technology to implement computer-aided diag-nosis at present.In this thesis, some issues of the Computer Aided Diagnosis(CAD) techno-logy of breast cancer microcalcifications are investigated mainly. The extraction of microcalcifications regions of interest, the localization of microcalcifications in mammograms and the detection system of microcalcifications cluster by pro-gramming with the software are implemented. The main contents are as follows:(1)In order to extract microcalcifications region of interest effectively in mammograms, making the operations of detection and classification carry out in it, the follow-up workload reduce, according to the differences in energy, gray and texture between microcalcification regions and normal regions, with the advantages of support vector machine on classification, combined with the detection method of microcalcifications based on the morphological analysis, an algorithm is presented to classify the regions which are whether micro-calcification region or normal region by Two-layer Support Vector Classifier to complete extraction. Experiments indicate that the algorithm has simple ope-ration , achieves high true positive detection rate 85.5% and low false positive 1.9%.(2)To detect microcalcification in the region of interest, according to the feature that microcalcifications are signals which are submerged in very high-frequency noises and low-frequency backgrounds, we can detect micro-calcifications in the region of interest using the Subtract in the space domain, removal the major low-frequency backgrounds, and detect it using the wavelet transform multiresolution analysis in the frequency domain, removal a part of low-frequency backgrounds and very high-frequency noise. We operate the logical AND operation with the result of wavelet transform and the result of the Subtract, this can removal the low-frequency backgrounds and very high-frequency noise. Eventually we realize the detection of microcalcifications.(3)Based on the algorithm research of microcalcifications cluster detection technology in mammograms, we programm using the visual programming environment of Visual C++6.0 and calling some Matlab toolbox functions, and implement the microcalcification cluster detection system. The system includes three modules, auxiliary function, extraction of microcalcification region of interest and detection in region of interest. The system can implement the corres-ponding function.