In many countries, breast cancer causes death for a high number of women. If doctors Detects the disease early using mammograms, they can treat patients more successfully and more women survive but medical professionals find it difficult to read mammograms because the images contain random visual interference, have small differences between light and dark areas plus people make errors. These factors cause results that incorrectly show a disease is present or incorrectly show a disease is absent. For this study, a new method that uses a convolutional neural network (CNN) with improved edges to find and categorize breast cancer in mammogram images was presented. To make the image quality better, the method uses multiple steps such as changing the size, making the noise smaller, making data values standard, making the data set more diverse and using Contrast Limited Adaptive Histogram Equalization (CLAHE) to prepare the pictures to make the contrast higher. By using edge detection methods, like Canny edge detection & Harris corner detection, the system shows the borders of lesions but also changes in structure more clearly before the CNN identifies specific features. The model labels mammogram images as either not cancerous or cancerous. To measure how well the model works, the researchers used metrics for accuracy, precision, recall and the F1-score. As the results showed, the CNN with improved edges identified categories more correctly and made fewer errors than traditional CNN methods. This model is a tool that uses computers to help find breast cancer early as well as it is helpful for radiologists when they make clinical decisions.
Efunbote et al. (2026) studied this question.