Kidney Diseases Classification using Hybrid Transfer-Learning DenseNet201-Based and Random Forest Classifier

There are several disease kinds in global populations that may be related to human lifestyles, social, genetic, economic, and other factors the nature of country they live in. Most recent studies have focused on investigating prevalent diseases spread population order minimize mortality risks, choose best method for treatment, improve community healthcare. Kidney is one most widespread health problems modern society. This study focuses kidney stones, cysts, tumors, three common types renal illness, using a dataset 12,446 CT urogram whole abdomen images, aiming move toward an AI-based diagnosis system while contributing wider field artificial intelligence research. In this study, hybrid technique used by utilizing both pre-train models feature extraction classification machine learning algorithms task image diagnosis. The pre-trained model Densenet-201 model. As well as Random Forest classification, Densenet-201-Random-Forest approach has outperformed many previous studies, having accuracy rate 99.719 percent.