Abstract: Patient outcomes depend on the early and precise identification of kidney cancers, yet manual diagnosis using CT images can be laborious and subjective. In this study, we introduce a novel hybrid framework for kidney tumor diagnosis that combines a convolutional neural network (CNN) with fuzzy logic-based picture enhancement. In order to enhance contrast between potential malignancies and healthy tissue, a fuzzy inference algorithm first modifies pixel intensities. A unique CNN trained on augmented kidney CT datasets then classifies the improved images. To expedite the process, we also create a web-based interface that allows doctors to submit CT scans, launch the hybrid pipeline, and evaluate prediction results (normal vs. tumor) and confidence scores following a secure login. Tests on publicly available CT datasets show that our approach outperforms baseline CNNs without fuzzy preprocessing, achieving robust recall and high accuracy (≈98–99%) for tumor instances. By emphasizing questionable areas and decreasing oversight, the suggested system has the potential to help radiologists.

keywords: CT imaging, fuzzy logic, deep learning, convolutional neural networks, and kidney tumor detection.

| DOI: 10.17148/IARJSET.2025125204