Breast Cancer Detection in Mammograms Using VGG19

Breast cancer is listed as one of the leading deadly diseases in women, therefore early diagnosis plays an important role in further increasing the probability of survival. This article presents VGG19-based deep learning to classify mammograms in four groups: Normal, Cancer, benign, and benign with call-back. Pre-processing techniques such as resizing the image, augmentation of the image, and transfer learning are used to enhance both the accuracy of the model and also minimize model overfitting. The VGG19 classifier architecture is adopted, fine-tuned, and evaluated along with the labeled mammogram datasets, achieving a test accuracy of 91 percent and computational AUC better than 0.90 for all classes. The method provides a a great sensitivity and specificity performance analysis, making it suitable for clinical use. It represents an affordable and scalable solution to all these concerns with the dataset and the scarce environment that will, therefore, open all the gates to early improved detection of breast cancers.