Computed diffusion-weighted imaging using 3-T magnetic resonance imaging for prostate cancer diagnosis

To assess the efficacy of computed diffusion-weighted images (cDWIs) of b = 2000 s/mm(2) (cDWI(2000)) generated from DWIs of b = 0 and 1000 for prostate cancer (PCa) diagnosis in comparison with that of measured original DWIs of b = 1000 (mDWI(1000)) and b = 2000(mDWI(2000)) using 3-T MRI. Eighty patients who underwent a preoperative MRI examination, including T2WI and DWI (b = 0, 1000, 2000 s/mm(2)), were enrolled in this study. Four combinations of images, protocol A (T2WI alone), B (T2WI + mDWI(1000)), C (T2WI + mDWI(2000)) and D (T2WI + cDWI(2000)), were assessed for their diagnostic capability. Areas under the receiver operating characteristic curve (Az) and diagnostic performance were evaluated, as well as contrast ratios (CR) between cancerous and non-cancerous lesions for each DWI. The highest CR was obtained with cDWI(2000) (0.29 +/- 0.16). Sensitivity, specificity, accuracy, and Az of the protocols were: A: 66.3 %, 59.4 %, 63.0 %, 0.67; B: 82.6 %, 62.0 %, 72.5 %, 0.80; C: 84.1 %, 66.5 %, 75.5 %, 0.86; D: 83.2 %, 70.0 %, 76.6 %, and 0.84, respectively The specificities and accuracies of protocol C and D were significantly higher than those of protocol B (P < 0.05). cDWI(2000) appears to be more effective than mDWI(1000), and at least as effective as mDWI(2000) for PCa diagnosis.