Effect of Contrast Limited Adaptive Histogram Equalization (CLAHE) on Breast Cancer Detection Using Residual Network (ResNet)

Breast cancer is a leading cause of mortality among women worldwide. This situation emphasizes the critical need for early and accurate diagnosis. This study was conducted to investigate the impact of Contrast Limited Adaptive Histogram Equalization (CLAHE) as a preprocessing step to enhance the mammography image quality for improved classification using the ResNet-18 deep learning model. The dataset contains 745 original and 9685 augmented mammography images. As a result of binary classification, applying CLAHE did not affect the training accuracy which was obtained as 96 % but increased validation and test accuracies from 92 % to 95 % and 95 % to 97% respectively. On the other hand, the CLAHE application decreased computer processing time from 243 minutes to 109 minutes.