Detection and Localization of Renal Cancer Using Deep Learning

Renal cell carcinoma (RCC) is a common cancer that affects adults and accounts for around 90% of kidney malignancies. If not detected and treated early, RCC can become life-threatening. The dataset consists of 560 whole-slide images. Here, resection images are taken to train the convolutional neural network (CNN). Pretrained networks like AlexNet and ResNet models are used to train and are proposed for classifying the images, and a custom architecture is developed. Also, detection of cancer is to be done by the model developed. The objective is to develop a deep learning-based system that can accurately detect and classify renal cancer with high sensitivity. By using pretrained models, the accuracy, efficiency, and speed of renal cancer diagnosis are investigated. The accuracy of ResNet is 97.27% and AlexNet is 91.44%. In this work, detection and localization of the renal cancer are done using You Only Look Once (YOLO).