Rotation invariant detection of benign and malignant masses using PHT

The digital mammography plays a very important role in early detection of breast cancer. Expert radiologist visually searches mammograms for specific abnormalities. Automatic detection using computer based methods can improve accuracy of diagnosis significantly. The Polar Harmonic Transform is superior to Pseudo Zernike Moment and Zernike Moment based method in terms of kernel generation, numerical stability and easier implementation. In this paper we proposed a rotation and scale invariant method for digital mammogram medical image analysis using PHT. Scale and translation invariance is achieved by normalization of moment and then rotation invariance is obtained by PHT. A k- nearest neighbor classifier is employed for classify the digital mammograms. To test and evaluate of the method several set of digital mammogram was evaluated with different rotation in different noisy condition. The correct classification percentage is calculated with different rotational angles. Experimental results are compared using Polar sine Transform (PST), Polar Cosine Exponential Transform (PCET) and Polar cosine Transform (PCT). Among all PCT show superiority of method in comparison to PCET and PST.