A new wavelet algorithm to enhance and detect microcalcifications

• We have proposed a new thresholding technique applied over wavelet coefficients for mammogram enhancement.
• We have utilized Shannon entropy to find the best ‘t’ in the wavelet domain.
• We have utilized Tsallis entropy to find the best ‘t’ in the wavelet domain.
• The proposed technique has better FROC test with 96.5% true positives and 0.36 false positives.

Image restoration and enhancement are two diverse fields in image processing that often get more attention from researchers. Especially, the role of these fields in the medical fraternity in aiding the radiologist in saving lives has been immensely felt in the last decade. Microcalcification enhancement and detection is one such well-explored area of research. Numerous microcalcification detection algorithms can be found in the literature. Yet, surprisingly, there are few left over which need to be addressed. For example, the literature proves the efficacy of wavelet transform in handling mammograms. But the success rate depends on the quality of enhancement which in turn depends on a critical parameter called threshold. Very few studies have been found on this aspect. The essential idea of this research is to utilize entropy based concepts (Shannon and Tsallis) to compute the threshold from wavelet coefficients. The modified wavelet coefficients will enhance the mammogram and simplify it to a simple two class problem. Our proposed algorithms are tested and validated through two experiments on the MIAS and UCSF databases. A FROC test yielded 96.5% true positives and 0.36 false positives.