Automated Segmentation of Renal Cell Carcinoma in CT Images using Deep Learning Techniques

Renal cell carcinoma (RCC) is a type of kidney cancer that can be challenging to detect early. RCC poses significant challenges due to its local invasion, metastasis, and resistance to conventional therapies. Therefore, early detection and accurate diagnosis are paramount for optimizing patient outcomes and guiding appropriate therapeutic interventions. The study aims to segment tumors in renal cell carcinoma of CT images using deep learning-based segmentation such as ResNet118, UNet, Fuzzy UNET, and SegNet. The performance metrics are compared and evaluated using ground truth segmented images. Among these networks, ResNet emerged with remarkable segmentation results, showcasing Precision of 0.912, Recall (0.908), Accuracy (0.986), Dice coefficient (0.883), and Intersection over Union (IoU) of 0.834. The study’s results could lead to improved patient outcomes through more precise and automated tumor detection in CT scans. Thus, deep learning-based segmentation paves a way for effective identification of renal cell carcinoma in CT images.