Nephrolithcipher: Decoding Nephrolithiasis Through Deep Neural Networks

Kidney stone disease (nephrolithiasis) is a growing global health concern, often diagnosed using medical imaging modalities such as computed tomography (CT). Manual interpretation of CT images can be subjective and error-prone. This study presents a comparative analysis of two established convolutional neural network (CNN) architectures—VGG16 and MobileNetV2—for the classification of kidney-related abnormalities, including stones, cysts, and tumors, using CT scan images. The models were evaluated based on classification accuracy, processing time, and suitability for medical imaging tasks. Preprocessing steps, including resizing and segmentation, were applied to enhance model performance. Experimental results showed that VGG16 achieved an accuracy of 96%, while MobileNetV2 reached 94%, indicating the effectiveness of both models, with VGG16 slightly outperforming in accuracy and MobileNetV2 in computational efficiency. This analysis offers insight into the strengths and limitations of each model, aiding future research in selecting appropriate CNN architectures for kidney stone detection.