Deep-Learning-Based Image Preprocessing and Classification for Kidney Disease Detection in CT Scans

Kidney diseases are serious medical conditions that have life-threatening consequences. Because of their impact on patient health, they are an important area of medical research. Medical imaging coupled with deep learning methods and models have proven effective in diagnosing kidney diseases. In this study, we investigated the accuracy of image-processing methods, namely edge detection and contrast-limited adaptive histogram equalization (CLAHE), for the detection of normal kidneys, kidney cysts, kidney tumors, and kidney stones. The results suggest that CLAHE is the more suitable image-processing method when the ResNet50 deep learning architecture is used to detect and diagnose kidney diseases, with an accuracy of 92%, compared with 84% for edge detection.