FusionEffNet: An EfficientNet-Based Transfer Learning Model with Dual-Path Feature Fusion and AdamW Optimization for Multi-Class Kidney Disease Classification from CT Scans

Abstract

The detection of chronic disease is always a major challenge in health care sectors. Early and accurate detection can help in fast recovery of these types of diseases. Several renal diseases such as cysts, tumors and stones may come in this category and can save life by early detection. The modern computational methods like machine learning and deep learning with smart medical imaging techniques may assist in reliable diagnosis, enhancing early detection of kidney illnesses and reducing diagnostic errors in the healthcare industry. In this study, the combination of convolutional architecture and optimized transfer learning techniques can improve diagnostic accuracy of kidney related diseases. A large dataset of 20077 CT images is used in the study, and the result shows optimal accuracy related to multi class kidney diagnostic classes including cyst, tumor, stone, and normal using FusionEffNet. The result of the study will defenetly improve the diagnosis process and treatment process in healthcare sectors.