Organ-Specific CT Radiomics Signatures of Hepatic, Pancreatic, and Renal Parenchyma for Stratification of Glycaemic Status

Abstract Purpose To evaluate organ-specific computed tomography (CT) radiomics signatures of the liver, pancreas, and kidneys for stratification of glycaemic status and to determine differential radiomic behaviour across abdominal parenchymal organs. Methods In this prospective study, adult patients undergoing non-contrast abdominal CT were categorised into normoglycemia, prediabetes, and diabetes groups based on fasting plasma glucose and HbA1c. Regions of interest were manually segmented in hepatic, pancreatic, and renal parenchyma. First-order attenuation features and texture features derived from GLCM, GLRLM, and GLSZM matrices were extracted using IBSI-compliant software. Organ-wise radiomic signatures were analysed for group differences and correlation with glycaemic markers. Receiver operating characteristic (ROC) analysis evaluated organ-specific stratification performance. Results A total of 120 patients (mean age 52 ± 11 years) were included. Hepatic mean attenuation showed significant inverse correlation with fasting glucose (r = − 0.41, p < 0.001). Pancreatic entropy and renal gray-level non-uniformity demonstrated significant positive correlations with HbA1c (r = 0.36 and 0.32, respectively; p < 0.01). Organ-wise AUCs for discrimination of diabetes were: liver 0.81, pancreas 0.76, and kidney 0.72. Multi-organ fusion improved AUC to 0.87. Conclusion CT radiomics reveals distinct organ-specific signatures associated with glycaemic status. The liver demonstrates the strongest attenuation-based association, while the pancreas and the kidney exhibit texture-driven heterogeneity changes. Multi-organ phenotyping enhances metabolic stratification beyond single-organ assessment.