A robust approach for denoising and enhancement of mammographic images contaminated with high density impulse noise

The categorisation of breast lesions as either benign or malignant if done accurately; would greatly reduce the mortality rate due to breast cancer all around the world. But the process is very challenging as the noise particles are generally detected as false positives which can be minimised only by the selective enhancement of the features of the mammogram indicative of cancer. This paper presents a combined approach for the suppression of high density impulse noise followed by the contrast enhancement of mammographic breast lesions. The application of the proposed denoising method is done iteratively to effectively remove the impulse noise. The non-linear enhancement operator with multistate adaptive gain is then passed over the denoised image for mammographic feature enhancement. Results of simulation show a marked improvement in the restoration quality of the contaminated images, preserving the finer features at high noise densities requiring not more than three iterations. With the optimal tuning of parameters, the enhancement of the targeted region of interest (ROI) is obtained with reasonable suppression of the background.