Abstract
OBJECTIVE: To evaluate the predictive value of whole-tumor apparent diffusion coefficient (ADC) histogram parameters derived from MRI for assessing lymph node metastasis (LNM) in pancreatic ductal adenocarcinoma (PDAC). MATERIALS AND METHODS: Preoperative MRI and clinical data from 53 patients with pathologically confirmed PDAC were retrospectively analyzed. Patients were divided into two groups: LNM (n = 29) and non-LNM (NLNM, n = 24). ADC maps were generated from diffusion-weighted images acquired on a 3.0 T MRI scanner. Whole-tumor regions of interest were delineated in FireVoxel software to extract the full-volume ADC histogram parameters. A predictive model was developed and assessed using ROC analysis. RESULTS: All ADC histogram parameters except the coefficient of variation and kurtosis showed significant differences between LNM and NLNM groups (p < 0.05); the first-order ADC values of LNM were significantly lower than those of NLNM. Baseline clinical characteristics (age, sex, clinical symptoms, CA19-9 levels) and conventional MRI features (size and volume) did not differ significantly. The multi-parameter model, based on select ADC-derived metrics, achieved an AUC of 0.865, with 86.2% sensitivity and 75.0% specificity. CONCLUSION: Whole-tumor ADC histogram analysis provides a non-invasive and quantitative tool for preoperative prediction of lymph node metastasis in PDAC. The integrated multiparametric model demonstrates superior diagnostic performance compared with single-parameter analysis.