Morpological Calcifications Detection Algorithms On Neural Networks In Mammograms

Breast cancer is one of the most common malignant diseases. Early diagnosis and treatment of breast cancer is a key ingredient of any strategy designed to reduce breast cancer mortality. At present, Mammography is the first choice to diagnose breast cancer. Clustered microcalcifications are one of the most important indications of malignancy in mammograms. Unfortunately, it is hard to recognize calcifications with naked eyes even in the high resolution mammograms due to the small size and various shape of calcifications.With the rapid development of medical imaging modalities and computer technology, computer aided detection of microcalcifications in mammo- grams have been a hot resharch field for early diagnosis of breast cancer.In order to detect microcalcifications in mammograms efficiently, this paper presents an automatic detection system model which consists of three modules on the basis of previous research. Firstly, a method is presented to extract the breast areas automatically in mammograms,which not only suppresses the background areas, but also simplifies the following detection of calcifications. Secondly, multiresolution wavelet transform and ICA were used to describe the characteristics of calcifications. The classic BP neural network trained by the characteristics is applied to distinguish the ROIs which come from partitions of breast areas. Finally, an algorithm based on top-hat morphologic operator is developed to enhance the microcalcifications in ROIs accurately. Then using the extracted features such as gray, texture, contrast etc., microcalcifications in ROIs were detected effectively.To verify the effectiveness of the proposed detection method, the experiments were conducted on the real mammograms. The experimental results illustrate that the detection modules take on good practicability and robustness, which can effectively detect regions of interest and microcalcifications in mammograms. The algorithm improves the preformance of computer aided detection significantly.