ANALYSIS OF PRE- PROCESSING TECHNIQUES AND SEGMENTATION TECHNIQUES FOR THE BREAST CANCER MRI IMAGES

Abstract

Abstract. This work is carried out to examine different filtering techniques to improve the image and remove noise from breast cancer MRI images, which aids in the subsequent assessment of MRI images, for precise Breast cancer identification, to contrast the suggested approach with current filtering methods in order to determine the most effective filtering method. MRI Images normally have noises such as salt and pepper, speckle. Various filtering techniques, such as the Median filter, Wiener filter , Mean Filter and Gaussian Filters are used for these noisy images. Metrics such as PSNR, SSIM, MSE, and SNR are used to calculate the performance of various filters. The optimal filter is so chosen to eliminate noise from the MRI Breast Cancer images based on its performance parameters. The findings of this work demonstrate that the Median filter outperforms other filtering techniques in removing noise from MRI Breast Cancer images.  The objective of this work is to obtain a precise image of the breast region using the Image Segmentation approach. The optimal segmentation method will be demonstrated by comparing and analyzing several segmentation algorithms. The filtered image is segmented using different segmentation Algorithms such as Fuzzy C Means Clustering, Thresholding, OTSU’s   and U- Net segmentation algorithms. It is determined that, in comparison to alternative segmentations, OTSU's approach accurately segments the breast MRI images based on the experimental results.