Development of computer-aided detection and diagnosing system for incipient stage of breast cancer

摘要

We proposed a CAD system to support radiologists in detecting the incipient stage of breast cancer. Our system is comprised of two processes. The first part is the detection of clustered microcalcifications and the second is the diagnosing of their malignancy. The essence of the first part is the extension of filter bank based on wavelet analysis. By using the method, the shape of each microcalcification and the relative density distribution of it are kept, so that we can evaluate accurately several features which may be used in the development of the second part. In this paper four features were used in order to evaluate the degree of malignancy. The performance of this system was estimated by using 100 mammograms including 53 abnormal images. The result shows the high performance of the detection and the diagnosis, also would prove a possibility that this system can support radiologists.