Objective: To assess the feasibility of computed diffusion-weighted imaging (cDWI) in comparison with directly acquired DWI for visualizing pancreatic adenocarcinomas.
Materials and methods: Patients with pancreatic adenocarcinoma underwent DWI at b-values of 0, 1000 (DVVI1000), 1500 (DWI1500) and 2000 (DWI2000) s/mm(2). From DWIs at b-values of 0 and 1000 s/mm(2), we generated cDWIs at b-values of 1500 (cDWI(1500)) and 2000 (cDWI(2000)) s/mm(2). DWI findings of pancreatic adenocarcinomas (clear hyperintensity; hyperintensity with an unclear distal border; and isointensity), the image quality and the tumor to pancreas contrast ratio (CR) were compared between directly acquired DWI and cDWI.
Results: Among the 63 included patients, clear hyperintense tumors were seen in 35 on DWI1000, 50 on DWI1500, 50 on cDWI(1500), 53 on DVVI2000 and 44 on cDWI(2000). Incidence of clear hyperintense tumors was significantly higher on cDWI(1500) than on DWI1000 = 0.013). There was no significant difference in the incidence of clear hyperintense tumors between DWI1500 and cDWI(1500) > 0.999), but a lower incidence was seen on cDWI(2000) than on DWI2000 (P = 0.028). Image quality was lower on cDWI than on DWI at b-values of 1500 (P = 0.002) and 2000 s/mm(2) (P < 0.001). The tumor to distal pancreas CR was significantly higher on cDWI(2000) than on cDWI(1500) (P < 0.001), and on cDWI(1500) than on DWI1000 (P < 0.001). The cDWI showed a significantly higher tumor to distal pancreas CR than DWI at b-values of 1500 (P = 0.004) and 2000 s/mm(2) (P < 0.001).
Conclusions: cDWI(1500) generated from b-values of 0 and 1000 s/mm(2) should be considered more effective than DWI1000 and at least as effective as DWI1500.