Magnetic Resonance Imaging at 14 Tesla of Prostate Biopsies Obtained after Prostatectomy

ABSTRACT This study aimed to investigate the feasibility of MR diffusion imaging of fresh prostate needle biopsy cores in a 14 Tesla vertical bore NMR system. Biopsies sampled from the index tumor and a presumed tumor‐free area of eight prostatectomy specimens underwent diffusion‐weighted imaging using conventional pulsed gradient spin echoes (PGSE) and stimulated echo acquisition mode (STEAM) with different gradient separation times between 10 and 400 ms at nominal b values ranging from 0 up to 5000 s/mm 2 . The apparent diffusion coefficient (ADC) was determined over the b ‐value range 0–1000 s/mm 2 . Kurtosis and biexponential models were fitted to measured signals at effective b values within the range 0–2000 and 0–5000 s/mm 2 , respectively. Moreover, monoexponential models were fitted over the same b ‐value ranges. After imaging, all biopsies underwent histopathological evaluation. For all diffusion gradient separation times evaluated, the ADC of histopathologically confirmed tumor and normal tissue was not significantly different ( p = 0.140 at 10 ms gradient separation time). Irrespective of the investigated tissue type, for an increasing gradient separation time up to 150 ms ADC decreased. For longer times up to 400 ms no further change was evident. For short diffusion gradient separation times, the Akaike information criterion (AIC) favored the biexponential model in almost all voxels, but for longer diffusion gradient separation times, in a majority of voxels the AIC favored the monoexponential model over either the kurtosis or biexponential model. Imaging of fresh prostate biopsies is feasible. At short diffusion times, in agreement with the presence of at least two distinct compartments, the signal decay is well described by a biexponential model. At long diffusion times, the ADC is lower, and a monoexponential model tends to be more appropriate, indicating the presence of exchange between compartments.