EXPLORE PUBLICATIONS BY COUNTRIES

Breast Cancer, Digital Mammography, Region of Interest.

Breast cancer is the most common type of cancer of women worldwide, but it can be cured if diagnosed at an early stage. Mammography is the main means of cancer screening and provides useful information on the signs of cancer, such as microcalcifications, masses, architectural distortion etc., which are not easy to distinguish due to certain defects in mammographic images, including low contrast, high noise, blurring and confusion. These challenges could be overcome by proposing a new preprocessing model. This work proposes a pre-processing method using different techniques and their combination in order to minimize the above-mentioned defects in mammographic images and make them usable for further processing. The different techniques range from filtering, thresholding, histogram, blur masking, morphological operations, thresholding and cropping. The aim is to put the mammographic images into a representation that will facilitate the detection of microcalcifications and the classification of healthy and cancerous images. Algorithms were developed and tested using the publicly available international database of the Mammographic Image Analysis Society (MIAS), which contains 322 samples. The results obtained on the regions of interest using four samples clearly show the background of the image and the objects (the part of the pectoral muscle and the suspicious area). These results show that much of the adipose tissue, fat mass and some of the features observed in the zoomed-in part of the image are significantly reduced. Furthermore, the results obtained in terms of visual quality compared to the literature show that they are better.