Abstract
BACKGROUND: To investigate the associations between the diffusion parameters obtained from multiple-b-values diffusion weighted imaging (DWI) of pancreatic ductal adenocarcinoma (PDAC) and the aggressiveness and local stage prediction, and assess the values of the quantitative parameters for the discrimination of tumors from healthy pancreas. METHODS: Fifty-one patients with surgical pathology-proven PDAC (size, 35 ± 12 mm) and fifty-seven healthy volunteers were enrolled. Diffusion parameters including monoexponential apparent diffusion coefficient (ADC(b) and ADC(total)) and biexponential intravoxel incoherent motion (IVIM) parameters (ADC(slow), ADC(fast) and f) based on 9 b-values (0 to 1000s/mm(2)) DWI were calculated for the lesions and the healthy pancreas. These parameters were compared by grades of differentiation, lymph node status, tumor stage and location. The diagnostic performances were calculated and compared by using the receiver operating characteristic curves (ROC) analyses. RESULTS: There was no statistically significant difference in ADC(b), ADC(total), ADC(slow), ADC(fast) or f between PDAC stage T1/T2 and stage T3/T4 or moderately differentiated versus poorly differentiated PDAC (p = 0.060-0.941). In addition, no significant differences were observed for the quantitative parameters between tumors located in the pancreatic head versus other pancreatic regions (p = 0.203-0.954) or between tumors with and without metastatic peri-pancreatic lymph nodes (p = 0.313-0.917). ADC(25-600), ADC(1000), ADC(total) and ADC(fast) were significantly lower for PDAC compared the healthy pancreas (all p < 0.05). ROC analyses showed the area under curve for ADC(20) was the largest (0.911) to distinguish PDAC from normal pancreas (cut-off value, 5.58 × 10(-3)mm(2)/s) and had the highest combined sensitivity (89.5%) and specificity (82.4%). CONCLUSIONS: Multiple-b-values DWI derived monoexponential and biexponential parameters of PDAC do not exhibit significance dependence on tumor grade or tumor characteristics. ADC(20) provided the best accuracy for differentiating PDAC from healthy pancreas in the study.