This study investigated whether MR-based radiomics can provide imaging biomarkers capable of distinguishing deficiency-type and excess-type Traditional Chinese Medicine (TCM) syndromes in patients with prostate cancer (PCa). A cohort of 121 men with PCa from two institutions was analyzed, with 84 allocated to a training set and 37 to an external validation set. According to TCM diagnostic criteria, patients were classified into deficiency or excess syndrome groups. Quantitative radiomic features were extracted from T2-weighted images (T2WI), diffusion-weighted sequences, and corresponding apparent diffusion coefficient (ADC) maps. Feature selection was performed in the training set using minimum redundancy maximum relevance followed by least absolute shrinkage and selection operator, yielding a radiomic signature for classification. Model performance was examined using receiver operating characteristic analyses and calibration assessments.
Across all three image types—T2WI, diffusion-weighted imaging, and ADC maps—patients presenting with excess syndromes showed significantly higher radiomic scores than those with deficiency syndromes. The T2WI, diffusion-weighted, and ADC models achieved areas under the ROC curve of 0.824, 0.824, and 0.847 in the training set, and 0.759, 0.750, and 0.809 in the validation set. Among these, the ADC-based model provided the strongest discriminatory capability, reaching accuracies of 0.788 in training and 0.778 in validation. Calibration results indicated good alignment between predicted radiomic outputs and actual TCM syndrome categories.
Radiomics derived from MR imaging offers a feasible, non-invasive strategy for differentiating TCM deficiency versus excess syndromes in PCa, with ADC-related features demonstrating the highest diagnostic value.
