Improved Classification of Kidney Lesions in CT Scans Using CNN with Attention Layers: Achieving High Accuracy and Performance

Kidney lesions are atypical areas of tissue damage or alteration in the kidneys that might indicate several illnesses, including tumors, cysts, or other pathological abnormalities. Precise identification and categorization of kidney abnormalities using medical imaging methods is essential for precise diagnosis and efficient treatment planning in nephrology. This paper introduces an innovative deep-learning method for precisely categorising CT kidney images. Our methodology seeks to automate the categorization procedure of kidney disorders, with a specific focus on differentiating between the Normal, Cyst, Tumor, and Stone categories. The CNN with attention layers utilizes its capacity to collect complex patterns and characteristics in the images, facilitating accurate identification and distinction of kidney abnormalities. The experimental findings on a complete CT kidney dataset exhibit exceptional performance. With a remarkable accuracy percentage of 97.98%, and average precision, detection, and F1 score of 98%. The accuracy and performance metrics attained demonstrate the efficacy and promise of the suggested method in aiding healthcare practitioners in the preliminary evaluation of kidney problems, ultimately resulting in enhanced diagnosis and treatment strategizing.