Kidney Stones, also known as nephrolithiasis, are a prevalent urological condition that causes severe discomfort and necessitates prompt identification and management. This study introduces a novel Deep Learning approach for automatically identifying kidney stones in computed tomography (CT) images. By utilizing advanced architectures such as MobileNetV2, VGG19 and InceptionV3, the proposed method enhances diagnostic accuracy, efficiency, and robustness. The methodology employs a comprehensive dataset of kidney stone images, which undergo preprocessing to enhance their quality and uniformity. The models are trained to identify kidney stones with varying dimensions, morphologies, and compositions, achieving remarkable accuracy rates of 94.87%, 95.15% and 94.77%. The study facilitates immediate, reliable clinical applications for early intervention, as evidenced by rapid processing times and extensive validation using independent datasets.