Kidney Disease Detection and Classification Based on CT Scan Images Using Deep Learning

Accurate detection and classification of kidney diseases are crucial within medical image processing for increased diagnostic precision and improved treatment outcomes. This research focuses on the automated detection and differentiation of five kidney conditions: normal, medullary sponge kidney, cyst, stone, and tumor, leveraging deep learning techniques. The detection and classification challenge lies in the subtle and overlapping features within medical images, particularly in the early stages of conditions. Additionally, it can incorporate medullary sponge kidney disease (MSK) as a new condition in the detection and classification process and improve the distinction between it and cysts. To address this, publicly available kidney image datasets were resized to 224×224 and 512×512 pixels and processed using transfer learning and fine-tuned DL models, including EfficientNetV2B0, DenseNet169, and ResNet50V2. Experimental results demonstrate that the EfficientNetV2B0 model, trained on 224×224 images and a balanced dataset, achieved a 97% testing accuracy with a precision, recall, and F1-score of 97% across all categories. ResNet50V2, applied to 512×512 images, achieved a comparable 96% testing accuracy. These findings highlight the effectiveness of deep learning in accurately detecting and classifying kidney pathologies, offering promising advancements in medical imaging and diagnostic tools.