Evaluating denoising performances of basic filters in the detection of microcalcifications on mammogram images

This research evaluates the denoising abilities of some image-processing filters used in facilitating the early detection of microcalcifications in breast tissues. The mean, median and Gaussian filters were employed to denoise mammogram images of microcalcification breast phantoms of various densities. The performances of the filters were assessed by evaluating the mean squared error (MSE), peak signal-to-noise ratio (PSNR), and signal-to-noise ratio (SNR). All experiments were carried out on MATLAB R2020a platform. The results revealed that the Gaussian filter recorded optimal performance in denoising images with all 3 types of added noises compared to the mean and median filters. The PSNR value of the heterogeneous phantom (PVAL/H) was superior to those of the less dense (PVAL/E), dense (PVAL), and extremely dense (PVAL/G) phantoms for all the tested filters. The results of this work agree with the high contrast recorded by the original image of PVAL/H.