Abstract: Kidney stones (nephrolithiasis) represent a widespread urological disorder affecting millions of individuals globally, frequently causing severe pain, obstruction of urine flow, urinary tract infections, and potentially irreversible renal damage when early detection is missed. Traditional diagnostic approaches rely on imaging modalities—principally ultrasound and computed tomography (CT)—which require expert radiological interpretation and may introduce delays or inter-observer variability. This study presents a Convolutional Neural Network (CNN)-based deep learning framework for the automated detection and classification of kidney stones from medical images. The proposed model integrates preprocessing pipelines, data augmentation strategies, hierarchical feature extraction, and rigorous performance evaluation using accuracy, sensitivity, and specificity metrics. Experimental results demonstrate that transfer learning architectures (VGG16, ResNet50, EfficientNet) significantly outperform classical machine learning classifiers and custom CNN designs, particularly when trained on CT imaging datasets. The system offers a cost-effective, scalable, and clinically integrable solution for diagnostic assistance, with the potential to reduce diagnosis time, minimize human error, and enhance patient outcomes.

Keywords: Kidney Stone Detection; Deep Learning; Convolutional Neural Networks; Medical Image Analysis; Transfer Learning; CT Scan; Automated Diagnosis; Nephrolithiasis

Downloads: | DOI: 10.17148/IARJSET.2026.13251