Development and Validation of a Renal Explainable AI Network (RenalXNet) for Detection and Differential Diagnosis of Renal Tumour using CT Radiomic Features

This study aims to develop and validate an explainable artificial intelligence based model RenalXNet for the detection and differential diagnosis of renal tumors using contrast enhanced CT radiomic features. The study is designed as a two phase observational diagnostic accuracy study.

Phase I involves retrospective analysis of CT images and clinical data from 350 patients to extract radiomic features and develop a predictive model for classification of benign and malignant renal tumors. The model performance will be evaluated using standard diagnostic metrics with histopathology as the reference standard.

Phase II involves prospective enrolment of 40 patients undergoing contrast enhanced CT for clinical validation of the developed model. The study also aims to assess model interpretability using explainable AI techniques to improve transparency and clinical applicability.

The hypothesis of the study is that the proposed RenalXNet model can accurately detect and differentiate renal tumors and provide clinically meaningful explanations to support radiological decision making.