Adaptive Threshold-Based Tumor Detection Algorithm For Mammograms Images

Breast cancer is without doubt the leading cancer among women, and it is one of the most damaging illnesses to females that should be periodically checked. Early detection of breast cancer can reduce the mortality caused by this disease by 95%. However, studies mention that up to 25% of tumors are missed by radiologists. In this paper, a tumor detection algorithm in mammogram images is developed by relying on simple calculations that are based on adaptive thresholding and tumor area size. Low complexity calculations will ease the implementation of the algorithm in embedded systems and in real-time detection. The proposed algorithm is used to detect the circular type of tumor and it is developed with a graphical user interface to ease the process of selecting mammogram images and changing settings of threshold values and the size of tumor area. Experimental results show the ability of the algorithm to successfully detect and differentiate circular tumors from normal and fatty breast tissue.