Tc-99m DMSA Radiomics in CKD: Phenotype-Specific Cortical Signatures and a Morphological Predictor of Renal Function...

Purpose: This study aims to evaluate the ability of radiomic features obtained from technetium-99m dimercaptosuccinic acid (Tc-99m DMSA) planar images to distinguish renal cortical uptake patterns among patients with chronic kidney disease (CKD). We also assessed the association between selected radiomic features and progressive renal function loss during follow-up. Methods: The study included a total of 185 patients: patients with Diabetes mellitus (DM) + hypertension (HTN) diagnosis (Group 1, n = 30), patients with HTN diagnosis alone (Group 2, n = 86), and patients with no history of DM or HTN who were followed for CKD (Group 3, n = 69). Intergroup comparisons were performed using the Kruskal-Wallis test with Bonferroni-corrected post hoc pairwise testing; the proportion of significantly different features was assessed using FDR correction. As a secondary exploratory analysis, the relationship between selected radiomic features and time to first observed ≥20% eGFR decline at follow-up was evaluated using univariate L2-penalised Cox proportional hazards regression with feature selection guided by the events-per-variable principle and model discrimination quantified using Harrell's C-index. Results: Intensity Kurtosis values showed a statistically significant difference among the groups: -0.11 (-0.31 to 0.12) for Group 1, -0.24 (-0.41 to -0.04) for Group 2, and -0.33 (-0.45 to -0.16) for Group 3 (p = 0.001). Mean Intensity values were found to be 60.66 (31.01-89.39) in Group 1 and 90.46 (72.87-106.34) in Group 3 (p < 0.001). Age, gender, and baseline eGFR did not differ between groups. Radiomic analysis revealed significant intergroup differences predominantly in intensity- and texture-based features, while morphological features showed more limited differentiation. In the secondary exploratory longitudinal analysis, Centre of Mass Shift was the only morphological feature significantly associated with time to first observed ≥20% eGFR decline at follow-up (HR per SD: 0.74; 95% CI: 0.58-0.94; p = 0.015; C-index: 0.57). Conclusions: Radiomic features from Tc-99m DMSA planar images reveal quantitative differences between clinically defined CKD subgroups even when cortical uptake appears visually indistinguishable. The threshold-specific association of Centre of Mass Shift with subsequent eGFR decline, beyond baseline renal function, suggests that DMSA radiomics may provide exploratory prognostic information that warrants prospective validation.