Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results

Purpose: To assess the feasibility of full diffusional kurtosis tensor imaging (DKI) in prostate MRI in clinical routine. Histopathological correlation was achieved by targeted biopsy. Materials and Methods: Thirty-one men were prospectively included in the study. Twenty-one were referred to our hospital with increased prostate specific antigen (PSA) values (>4 ng/ml) and suspicion of prostate cancer. The other 10 men were volunteers without any history of prostate disease. DKI applying diffusion gradients in 20 different spatial directions with four b-values (0, 300, 600, 1000 s/mm(2)) was performed additionally to standard functional prostate MRI. Region of interest (ROI)-based measurements were performed in all histopathologically verified lesions of every patient, as well as in the peripheral zone, and the central gland of each volunteer. Results: DKI showed a substantially better fit to the diffusion-weighted signal than the monoexponential apparent diffusion coefficient (ADC). Altogether, 29 lesions were biopsied in 14 different patients with the following results: Gleason score 3 + 3 = 6 (n = 1), 3 + 4 = 7 (n = 7), 4 + 3 = 7 (n = 6), 4 + 4 = 8 (n = 1), and 4 + 5 = 9 (n = 2), and prostatitis (n = 12). Values of axial (K-ax) and mean kurtosis (K-mean) were significantly different in the tumor (K-ax 1.78 +/- 0.39, K-mean 1.84 +/- 0.43) compared with the normal peripheral zone (K-ax 1.09 +/- 0.12, K-mean 1.16 +/- 0.13; p < 0.001) or the central gland (K-ax 1.40 +/- 0.12, K-mean 1.44 +/- 0.17; p = 0.01 respectively). There was a minor correlation between axial kurtosis (r = 0.19) and the Gleason score. Conclusion: Full DKI is feasible to utilize in a routine clinical setting. Although there is some overlap some DKI parameters can significantly distinguish prostate cancer from the central gland or the normal peripheral zone. Nevertheless, the additional value of DKI compared with conventional monoexponential ADC calculation remains questionable and requires further research. (C) 2014 Elsevier Inc. All rights reserved.