Fast fractal modeling of mammograms for microcalcifications detection
Clusters of microcalcifications in mammograms are an important early sign of breast cancer in women. Comparing with microcalcifications, the breast background tissues have high local self-similarity, which is the basic property of fractal objects. A fast fractal modeling method of mammograms for detecting the presence of microcalcifications is proposed in this paper. The conventional fractal modeling method consumes too much computation time. In the proposed method, the image is divided into shade (homogeneous) and non-shade blocks based on the dynamic range and only the non-shade blocks are modeled. Reducing the number of the processed blocks reduces the encoding time to 6.372% compared to the conventional modeling method. The modeled mammograms were investigated for microcalcifications detection and the results show that the sensitivity is 92% for 25 abnormal mammograms were obtained.