Early Breast Cancer Detection Among Women Using Deep Learning Algorithms: A Case Study of Oyo State, Nigeria

Abstract Breast cancer has remained one of the leading causes of death in Nigerian women, mostly due to late diagnosis and inadequate diagnostic resources. We investigate the feasibility of using deep learning for enhancing detection in Oyo State, Nigeria, through the application of CNNs as compared to the human diagnostic process. Hospital and diagnostic center imaging was used and analyzed with Residual Network-50, VGG16, and Efficient Network deep learning models. Human diagnosticians reported 75% accuracy, but the Residual Network-50 model showed superior detection results of 94% accuracy, 93% sensitivity, and 95% specificity. Efficient Network was equally competitive, showing a balanced accuracy of 92%, but required less computing power. The deep learning models, especially the Residual Network-50 and Efficient Network models, showed potential to enhance the accuracy in breast cancer detection, especially in resource-limited areas. Overcoming limitations of implementation will assist in bringing down the mortality rates and increase the well-being of women in Oyo State and in Nigeria in general.