High-Resolution Diffusion-Weighted Imaging of the Prostate

Transcription

1 Genitourinary Imaging Original Research Medved et al. High-Resolution DWI of the Prostate Genitourinary Imaging Original Research Milica Medved 1 Fatma N. Soylu-oy 1,2 Ibrahim Karademir 1,3 Ila Sethi 1,4 mbereen Yousuf 1 Gregory S. Karczmar 1 ytekin Oto 1 Medved M, Soylu-oy FN, Karademir I, et al. Keywords: high-resolution diffusion-weighted imaging (DWI), MRI, prostate cancer DOI: /JR Received pril 10, 2013; accepted after revision ugust 29, Supported by a University of hicago ancer enter Support Grant. 1 Department of Radiology, University of hicago, 5841 S Maryland ve, M 2026, hicago, IL ddress correspondence to M. Medved (mmedved@uchicago.edu). 2 Present address: Department of Radiology, Mehmet Training and Research Hospital, Istanbul, Turkey. 3 Present address: Department of Radiology, Eskisehir Military Hospital, Eskisehir, Turkey. 4 Present address: Department of Radiology, Louis Weiss Memorial Hospital, hicago, IL. JR 2014; 203: X/14/ merican Roentgen Ray Society High-Resolution Diffusion-Weighted Imaging of the Prostate OJETIVE. The purpose of this study was to evaluate the effect of increasing the spatial resolution of the prostate DWI protocol on image quality and lesion conspicuity. SUJETS ND METHODS. Twenty-nine patients with biopsy-proven prostate cancer undergoing MRI examinations were imaged with two diffusion-weighted imaging (DWI) protocols: current standard clinical protocol (6.7 mm 3 voxels) and a new high-resolution protocol (3.1 mm 3 voxels). Diffusion-weighted images were independently and subjectively scored on lesion conspicuity, internal architecture definition, and overall image quality by two radiologists. verage apparent diffusion coefficient (D) values were measured in normal tissue and cancerous lesions on both sequences. Reader scores and D and contrast values were compared between the two protocols. ancer D values were correlated with Gleason scores. RESULTS. The signal-to-noise ratio of the new high-resolution DWI protocol was 40% lower than that of the standard protocol. The reader scores were higher by 0.73 (range, ) grades, or 19% (range, 7 32%), on average, for the new protocol, indicating better image quality. The average D values were 8% higher with the new protocol, with D contrast values between cancer and normal prostate unchanged. There was marginally significant correlation of cancer D values with Gleason scores (p = 0.05, r 0.36). ONLUSION. We showed that for DWI of the prostate at 3 7 mm 3 voxel sizes the benefits of higher spatial resolution outweigh the effects of reduced signal-to-noise and contrast-to-noise ratios, potentially improving the sensitivity to small or sparse prostate cancers. Radiologists can consider using higher-spatial-resolution DWI sequences in their practices. I n recent years, MRI is increasingly being used in prostate cancer management, with applications ranging from diagnosing and staging to biopsy guidance [1 4]. T2-weighted imaging is increasingly being supplemented with functional imaging, further improving the utility of prostate MRI [2 6]. For example, diffusion-weighted imaging (DWI) has been shown to significantly improve the sensitivity of prostate MRI for prostate cancer over T2-weighted imaging alone [2, 7 9], and the newly published European Society of Urogenital Radiology prostate MR guidelines strongly endorse multiparametric MRI and stress the value of DWI [10]. However, there is room for improvement because apparent diffusion coefficient (D) values can have significant overlap between cancerous and noncancerous tissue, limiting specificity [2]. One of the factors contributing to the overlap of D values between cancerous and noncancerous tissue is likely the high hetero- geneity of prostate tumors. Langer et al. [11] found that 36% of tumors in the peripheral zone had pronounced intermixing of normal tissue within the malignant glands, surpassing the malignant glands in surface area. They labeled such tumors sparse. This intermixing occurs to varying degrees, with lesions in which normal glands are less frequent labeled dense, but the heterogeneity is typical and affects the D values and other characteristics measured by MRI. The distributions of D and T2 values measured in sparse lesions were similar to those observed in the normal tissue, whereas dense lesions exhibited lower D and T2 values [11]. Thus, it appears that higher spatial resolution imaging, which could allow better delineation of normal and cancerous tissue, would reduce the overlap of the MRI-measured quantities, including the D, and aid the sensitivity and specificity of prostate MRI. Good anatomic imaging in which lesions are clearly delineated from the healthy tissue can aid in JR:203, July

2 the diagnosis as well as improve the performance of MRI-guided biopsy. However, current DWI protocols at 3 T typically with a 2- to 3-mm in-plane resolution in 3-mm-thick slices [12 16] do not provide these benefits and higher-resolution sequences are needed. The optimal set of parameters for 3-T DWI is not clear because there is a trade-off between the spatial resolution, number of b values used, and signal-to-noise (SNR) and contrastto-noise (NR) ratios of the acquired images. This is an area of ongoing effort. We have tested a new high-spatial-resolution DWI prostate imaging protocol at 3 T, with the goal of achieving better anatomic imaging without compromising the functional information. This protocol was compared with the state-of-the-art protocol currently available on Philips Healthcare chieva 3T TX scanners. In comparison with the current clinical protocol, the voxel volume was halved, which in turn can reduce the SNR of the obtained images, depending on other parameters such as the number of b values. Thus, it is not a priori clear that the new high-resolution protocol would improve image quality, and we tested this assumption in a patient population. If better delineation of prostate lesions were achieved, this could potentially improve the sensitivity and specificity of prostate MRI. Subjects and Methods Patient Population In our clinic, patients with prostate cancer confirmed on biopsy are referred for MRI for evaluation of extent of disease. Subjects for this prospective study were recruited from this population under a protocol approved by the institutional research board and after informed consent was provided. The inclusion criteria included biopsy-confirmed prostate cancer and no prior treatment. HIP privacy rules were followed. Twenty-nine men were recruited sequentially for the study, with a median age of 63 years (range, years) and mean prostate-specific antigen (PS) level of 10.0 ng/ml (range, ng/ml). ll patients underwent transrectal ultrasound (TRUS)-guided biopsy and were found positive for prostate cancer in at least one core. In each TRUS-guided biopsy, at least one core was obtained from each sextant of the prostate (right base, left base, right middle, left middle, right apex, left apex) and separately evaluated and reported by the pathologist. The authors of the study did not perform the TRUS-guided biopsies; pathology reports were used for correlation with MRI. The average delay between biopsy and the MRI examination was 6.7 weeks. Medved et al. TLE 1: MRI Sequence Parameters for Diffusion-Weighted Imaging (DWI) Protocols Parameter MRI Protocol We used a 3-T scanner (chieva TX, Philips Healthcare). combination of a single-channel endorectal coil and either a six-channel cardiac or a 16-channel torso coil was used. When the torso coil was used, only the lower eight elements were active. The endorectal coil was inflated with approximately 60 ml of barium. Peristalsis was suppressed with an intramuscular injection of 1 mg of glucagon before the examination. Our standard clinical prostate imaging protocol was followed, and in addition, a higher-resolution DWI sequence was included for comparison with the Standard linical DWI Protocol New High-Resolution In-plane resolution (mm) Slice thickness (mm) 3 2 No. of slices FOV (mm) SENSE factor 2 2 Partial Fourier Fat suppression SPIR SPIR b values (s/mm 2 ) 0, 50, 150, 990, , 1200 (n = 18); 0, 700 (n = 11) No. of averages for each b value in each scan 1, 1, 1, 2, 3 1, 3 No. of averages for repeated scans 4 8 TR/TE (ms) 4995/ /66 (n = 18); 4653/58 (n = 11) Scan duration (min:s) 7:24 6:32 (n = 18); 6:17 (n = 11) Note SENSE = sensitivity encoding, SPIR = spectral attenuated inversion recovery. current clinical DWI sequence. Thus, the imaging protocol included the following sequences: coronal and sagittal T2-weighted fast spin-echo (TR/TE, 4750/115), axial T2-weighted dual-echo fast spin-echo (TR/TE, 3670/115), axial T1-weighted spoiled spin-echo (TR/ TE, 575/16), axial DWI (current clinical protocol [Table 1]), axial DWI (new high-resolution protocol [Table 1]), axial T1-weighted whole-pelvis survey with breathhold unenhanced (TR/TE, 3/1.4; flip angle, 10 ), axial T1-wieghted dynamic contrast-enhanced spoiled gradient-echo (TR/TE, 3/1.6; flip angle, 15 ), and axial T1-weighted whole-pelvis survey with breath-hold contrast-enhanced (TR/TE, 3/1.4; flip angle, 10 ). TLE 2: verage Image Quality Scores and omparison of Two Diffusion-Weighted Imaging (DWI) Protocols Parameter DWI Standard linical Protocol verage Scores ± SD DWI New High-Resolution Protocol Reader 1 Lesion conspicuity 4.0 ± ± Internal architecture definition 3.6 ± ± 0.4 < Overall image quality 3.9 ± ± 0.0 < Reader 2 Lesion conspicuity 4.1 ± ± a Internal architecture definition 3.7 ± ± 0.5 < Overall image quality 4.0 ± ± verage scores Lesion conspicuity 4.0 ± ± Internal architecture definition 3.6 ± ± 0.4 < Overall image quality 4.0 ± ± 0.3 < a Not statistically significant. p 86 JR:203, July 2014

3 High-Resolution DWI of the Prostate Diffusion-Weighted Imaging Protocols Our new high-resolution DWI sequence further increases the in-plane resolution and decreases the slice thickness, halving the voxel size relative to the standard clinical DWI protocol. oth DWI protocols were based on a sensitivity encoding accelerated singleshot echo-planar imaging sequence, and the imaging parameters for the two DWI protocols are listed in Table 1. Most important, the imaging times were comparable, which was achieved by simultaneously increasing the spatial resolution and reducing the number of b values scanned. In the new high-resolution protocol, 18 patients were scanned with the single nonzero b value of 1200 s/mm 2 (selected to be within the range of b values of the standard clinical protocol) and 11 were scanned with the single nonzero b value of 700 s/mm 2 (selected to optimize SNR). The number of accrued patients was not sufficient to allow detailed comparison between the high b value of 1200 s/mm 2 and high b value of 700 s/mm 2 groups; this is not the focus of the current work. For the purposes of this article, all high-spatial-resolution studies were treated as a single group. In three patients scanned with the high b value of 1200 s/mm 2 and in three patients scanned with the high b value of 700 s/mm 2 under the new high-resolution protocol, both DWI protocols were repeated so that SNR could be evaluated. Image Processing Diffusion-weighted images were processed either on the imaging console using Philips Healthcare software, or on a Philips Healthcare imaging workstation to produce D maps, which were then uploaded to our picture archiving and communication system (PS). The radiologists defined regions of interest (ROIs) outlining the cancerous lesions in representative slices. To do this, in the sextants found positive for cancer on biopsy, low-intensity T2-weighted regions and low D regions were correlated and identified as cancerous lesions. Similarly, normal tissue ROIs were selected in the peripheral zone by correlating uniform high-intensity T2-weighted regions and higher D regions. verage D values were calculated over the ROIs from values reported in the D maps. To calculate SNR, ROIs with approximately uniform high D values were identified on two corresponding slices imaged with the standard clinical and the new high-resolution protocols. The SNR was calculated as the ratio of the average signal intensity in the ROI and the SD in the ROI of the difference between repeated scans of the same DWI protocol. The repeated scans were scaled to the same average value across the entire prostate. ecause the SNR varies across the prostate, the ratio of SNRs in the new high-resolution relative to the current clinical protocol was calculated for each ROI and the average value is reported. Image Evaluation The MR images were reviewed with biopsy results, and cancer ROIs were identified by an experienced radiologist with 13 years of experience reading abdominal MRI examinations (reader 1). The radiologist identified cancer lesions as regions with low D and hypointense on T2-weighted images in the sextants that were reported positive on biopsy. The second radiologist, with 4 years of experience reading abdominal MRI examinations (reader 2) defined the normal tissue ROIs within the peripheral zone. The D images of identified lesions were graded by both readers, separately and independently on a 5-point scale (1, very poor; 2, poor; 3, average; 4, good; and 5, excellent) for lesion conspicuity, definition of internal architecture, and overall image quality. The lesions were graded for the two DWI protocols independently. Statistical nalysis The image quality scores for the two DWI protocols were compared using the Wilcoxon signed rank test for each reader separately and for scores averaged over the two readers. No correction for multiple comparisons was done because the scores in different categories are highly correlated. For comparison between the two DWI protocols, we used the Wilcoxon signed rank test to compare the D values and the ratios of D values between lesion and normal tissue ROIs and to compare the ratios of average D values in lesions to those of normal tissue (contrast). The Pearson correlation coefficient r was calculated for D values versus the Gleason score of the lesions. Results Patient Population For four patients, the images were not usable because of the following technical problems: wrong high-resolution DWI sequence was prescribed (n = 1), high-resolution DWI was not archived (n = 1), and all DWI studies had severe spatial distortion secondary to faulty endorectal coil (n = 2). Twenty-five examinations were available for image quality evaluation. Seven patients had missing or insufficiently detailed pathology reports. Thirty-one lesions located in the peripheral zone were identified in 18 patients, with a mean Gleason score of 7.0 (range, 6 9). The median lesion size was 13 mm (range, 7 27 mm). Fig year-old man with biopsy-proven prostate cancer (prostate-specific antigen = 4.08 ng/ml)., T2-weighted () and axial standard clinical () (b = 0, 50, 150, 990, and 1500 s/mm 2 ) and high-resolution () (b = 0 and 1200 s/mm 2 ) diffusion-weighted images through prostate show hypointense mass in left posterolateral apical peripheral zone that appears dark on both standard clinical and new high-resolution protocol apparent diffusion coefficient map images (arrow, ) with Gleason score of 7 (3 + 4). However, borders of lesion can be better seen under new protocol and appear more blurry on standard clinical protocol image. JR:203, July

4 Medved et al. Fig year-old man with biopsy-proven prostate cancer (prostate-specific antigen = 3.90 ng/ml)., T2-weighted () and axial standard clinical () (b = 0, 50, 150, 990, and 1500 s/mm 2 ) and high-resolution () (b = 0 and 1200 s/mm 2 ) diffusion-weighted images through prostate show prostate cancer in mid posterolateral aspect of right peripheral zone on T2-weighted image (arrows, ) with Gleason score of 6 (3 + 3) that is dark on apparent diffusion coefficient maps acquired under both protocols, but fine morphologic details of lesion (arrows, ) are better appreciated under new protocol. Fig year-old man with biopsy-proven prostate cancer (prostate-specific antigen = 5.40 ng/ml)., T2-weighted () and axial standard clinical () (b = 0, 50, 150, 990, and 1500 s/mm 2 ) and high-resolution () (b = 0 and 1200 s/mm 2 ) diffusion-weighted images through prostate show large prostate cancer in right peripheral zone (arrow, ) (Gleason score, 9 (4 + 5). Mass can be seen on both apparent diffusion coefficient (D) maps; however, note depiction of ducts in left peripheral zone on new high-resolution protocol D map image. Standard clinical protocol D map image is blurry and ducts cannot be delineated. Fig year-old man with biopsy-proven prostate cancer (prostate-specific antigen = 2.10 ng/ml)., T2-weighted () and axial standard clinical () (b = 0, 50, 150, 990, and 1500 s/mm 2 ) and high-resolution () (b = 0 and 1200 s/mm 2 ) diffusion-weighted images through prostate show prostate cancer in left posterolateral peripheral zone (arrow, ) with Gleason score of 7 (3 + 4) that can be better delineated with improved morphologic detail and conspicuity on new high-resolution protocol apparent diffusion coefficient (D) map image compared with standard clinical protocol D image. 88 JR:203, July 2014

5 High-Resolution DWI of the Prostate Qualitative nalysis Overall, the new protocol provided better boundary and detail delineation, including ducts. This is illustrated in Figures 1 4. The scores for lesion conspicuity, internal architecture definition, and overall image quality averaged over all lesions are summarized in Table 2 for each reader separately and averaged over the two readers. Statistically significant differences in the scores between the standard clinical and the new high-resolution protocol were found in all categories except lesion conspicuity for reader 2. In all cases (except lesion conspicuity for reader 2), the p values were 0.01 or less, and the new high-resolution DWI protocol scored higher, indicating better image quality. The scores were higher by 0.73 grades (range, grades), or 19% (range, 7 32%) on average. Quantitative nalysis The average relative SNR ± SD (the ratio of SNR in the standard clinical protocol to that in the new high-resolution protocol), was 0.50 ± 0.17 for the high b value of 1200 s/mm 2 and 0.74 ± 0.30 for the high b value of 700 s/mm 2, with the average relative SNR of 0.60 ± 0.26 for all examinations combined. Twenty-one ROIs in six patients were analyzed for the SNR calculation. The D values measured in lesions and normal tissue and their relevant ratios are summarized in Table 3. oth the standard clinical and the new high-resolution protocols distinguished well between lesions and normal peripheral zone tissue (p < 0.001). The D values measured with the new high-resolution protocol were higher for both lesions and normal tissue (p < 0.001) by approximately 8% in both cases. The ratio of D values in lesions to those in normal tissue (D contrast) was not significantly different (p = 0.36) between the two imaging protocols. orrelation With Gleason Score Only a weak and marginally significant (p = 0.05) correlation was found between the D values and the Gleason scores of the lesions for both the standard clinical (r = 0.35) and the new high-resolution (r = 0.37) imaging protocols. Discussion The new higher-spatial-resolution DWI protocol reduces the voxel size by a half and results in decreased SNR (by ~ 40%), as expected. Nevertheless, we found that the new protocol improves lesion delineation and lesion contrast. Thus, the higher-resolution DWI protocol could be useful for increasing the sensitivity for prostate cancer detection on MRI. The current clinical protocol already uses voxels that are about one half the size of those typically used at 3-T DWI. The image quality scores increased after a further reduction by a factor of two in voxel size (with the new protocol), which indicates that we have not reached the point of diminishing returns with increasing spatial resolution. The SNR in prostate DWI is not exhausted at the 3.1 mm 3 voxel size used in this study, and even higher spatial resolution e.g., in 1-mm-thick slices may bring additional diagnostic benefits. The subjective lesion conspicuity (contrast) scores were higher in the new protocol, even though the objective measure, the ratio of D values in lesions and normal tissue, did not improve. likely explanation is that although the average contrast remained the same, margin definition improved in the new higher-resolution protocol that is, the edges of the lesion are less blurred by the partial volume effect. Our median lesion size was 10 image voxels across, and this could be the reason we did not measure an increase in the objective measure of lesion contrast. The volume averaging effect would be more pronounced for smaller lesions, on the order of a few voxels across, in which there is very little tissue that is far away from the boundaries. Such smaller lesions could likely be depicted with higher contrast using the new protocol, which would aid detection. The partial volume effects are of particular interest for detection of sparse lesions, which may account for more than one third of all lesions [11]. In such lesions, the measured T2 and D values are intermediate to the values present in normal and cancerous tissue, and the lesions are thus less conspicuous [11]. Our new high-resolution DWI protocol, or perhaps one with even higher, e.g., 1 mm 3 spatial resolution, would increase the conspicuity of sparse lesions. This proposal would have to be tested experimentally because the reduction in NR might offset the improvements in the volume averaging effect. nother application in which partial volume effect reduction would be beneficial is the histogram analysis of the D values for cancer detection or characterization. lthough this technique has not been implemented in prostate imaging, it is already being explored in the cervix [17, 18] and in the brain [19, 20]. The new protocol measures a single rather than multiple nonzero b values. This precludes higher-precision measurements of D obtained by using biexponential fits to the signal decay over time data and accounting for the perfusion component of signal loss. Such measurements are inherently time-consuming and likely cannot be combined with high spatial resolution in clinically feasible times. Such sequences with multiple b values may be better suited for characterization of larger or already detected lesions, e.g., during therapy monitoring, whereas high-spatial-resolution protocols may be more useful for detection of small or sparse cancers. The D values in both the normal tissue and cancer were 8% higher on average in the new higher-resolution protocol than in the current standard clinical protocol. When low b values are not acquired, it is not possible to account for the contribution of perfusion, which can artificially increase the D. This did not affect the lesion contrast and thus did not affect the sensitivity for lesion detection. However, it may have implications for studies in which an absolute measurement of D is required. The average interval between the biopsy and MRI examination is shorter than the rec- TLE 3: verages of pparent Diffusion oefficient (D) Values and Ratios Parameter Standard linical DWI Protocol verage Values ± SD New High-Resolution DWI Protocol Ratio p (Imaging Protocols) D values Normal tissue (D N ) 1657 ± 232 (10 3 mm 2 /s) 1788 ± 294 (10 3 mm 2 /s) 1.08 ± 0.11 < a Lesion (D L ) 844 ± 192 (10 3 mm 2 /s) 910 ± 212 (10 3 mm 2 /s) 1.08 ± 0.11 < a p (normal tissue to lesion) < a < a 0.13 D contrast (D L / D N ) 0.53 ± ± a Statistically significant. JR:203, July

6 Medved et al. ommended 8 weeks [21]. lthough this may have biased the measured T2 and D values, the effect would be similar in the two studied protocols and is not likely to affect the results of this study. Furthermore, given the time course of the clinical diagnostic process, this is an issue that would be difficult to avoid in any MRI study of newly diagnosed prostate cancer. Our result for Pearson correlation coefficient r between the D and Gleason score was modest and marginally significant. It is possible that a higher number of lesions would lead to higher statistical significance. However, the magnitude itself is within the range of published values, which range from 0.60 (p = 0.03) to 0.28 (p = 0.11) [22 24]. limitation of the study is the lack of correlation with a prostatectomy specimen. ll of the patients were referred for MRI to evaluate the extent of disease that was detected on TRUS-guided biopsy, and thus all had intact prostates. This can be acceptable because we did not seek to establish the sensitivity of the new protocol but characterized its performance on known lesions. nother limitation is that our subjects underwent scanning with two different high b values, which might have increased somewhat the variability of the measured D values. Finally, the new protocol covers a smaller volume, which may become an issue in cases of very large prostates. The acquisition volume thickness (48 mm) was sufficient to cover the entire prostate in all of our subjects, but if larger anatomic coverage is desired, the scanning times will be extended. This can be offset by using a smaller number of averages because we have not exhausted the SNR in implementing the new protocol. In conclusion, the new high-resolution DWI protocol provided significant benefits compared with the current clinical DWI protocol. The higher spatial resolution resulted in better lesion conspicuity, better defined internal architecture, and better overall image quality. The better margin definition likely results from reduced partial volume effects, which could increase sensitivity to small and sparse lesions and allow novel applications, such as D histogram analysis. The pursuit of even higher-resolution DWI protocols is a meaningful goal, although at this stage it precludes multiple b value measurements that could take into account the perfusion effects. Given the promising results of this preliminary study, further evaluation of the incorporation of this approach into routine clinical protocol is warranted. References 1. igner F, Pallwein L, Pelzer, et al. Value of magnetic resonance imaging in prostate cancer diagnosis. World J Urol 2007; 25: hoi YJ, Kim JK, Kim N, Kim KW, hoi EK, ho KS. Functional MR imaging of prostate cancer. RadioGraphics 2007; 27: Somford DM, Fütterer JJ, Hambrock T, arentsz JO. Diffusion and perfusion MR imaging of the prostate. Magn Reson Imaging lin N m 2008; 16: Kim JK, Jang YJ, ho G. Multidisciplinary functional MR imaging for prostate cancer. Korean J Radiol 2009; 10: Tan H, Wang J, Kundra V. Diffusion weighted imaging in prostate cancer. Eur Radiol 2011; 21: Panebianco V, Sciarra, Marcantonio, et al. onventional imaging and multiparametric magnetic resonance (MRI, MRS, DWI, MRP) in the diagnosis of prostate cancer. Q J Nucl Med Mol Imaging 2012; 56: Shimofusa R, Fujimoto H, kamata H, et al. Diffusion-weighted imaging of prostate cancer. J omput ssist Tomogr 2005; 29: Tan H, Wei W, Johnson V, Kundra V. Diffusionweighted MRI in the detection of prostate cancer: meta-analysis. JR 2012; 199: Wu LM, Xu JR, Ye YQ, Lu Q, Hu JN. The clinical value of diffusion-weighted imaging in combination with T2-weighted imaging in diagnosing prostate carcinoma: a systematic review and meta-analysis. JR 2012; 199: arentsz JO, Richenberg J, lements R, et al. ESUR prostate MR guidelines Eur Radiol 2012; 22: Langer DL, van der Kwast TH, Evans J, et al. Intermixed normal tissue within prostate cancer: effect on MR imaging measurements of apparent diffusion coefficient and T2-sparse versus dense cancers. Radiology 2008; 249: Park SY, Kim K, Park K, Lee HM, Lee KS. Prediction of biochemical recurrence following radical prostatectomy in men with prostate cancer by diffusion-weighted magnetic resonance imaging: initial results. Eur Radiol 2011; 21: Metens T, Miranda D, bsil J, Matos. What is the optimal b value in diffusion-weighted MR imaging to depict prostate cancer at 3T? Eur Radiol 2012; 22: Kitajima K, Kaji Y, Fukabori Y, Yoshida K, Suganuma N, Sugimura K. Prostate cancer detection with 3 T MRI: comparison of diffusion-weighted imaging and dynamic contrast-enhanced MRI in combination with T2-weighted imaging. J Magn Reson Imaging 2010; 31: Kim K, Park K, Park W, Kim SS. Prostate MR imaging at 3T using a phased-arrayed coil in predicting locally recurrent prostate cancer after radiation therapy: preliminary experience. bdom Imaging 2010; 35: Kim K, Park K, Kim. Diffusion-weighted MRI at 3 T for the evaluation of prostate cancer. JR 2010; 194: Downey K, Riches SF, Morgan V, et al. Relationship between imaging biomarkers of stage I cervical cancer and poor-prognosis histologic features: quantitative histogram analysis of diffusionweighted MR images. JR 2013; 200: Rosenkrantz. Histogram-based apparent diffusion coefficient analysis: an emerging tool for cervical cancer characterization? JR 2013; 200: Nowosielski M, Recheis W, Goebel G, et al. D histograms predict response to anti-angiogenic therapy in patients with recurrent high-grade glioma. Neuroradiology 2011; 53: Pope W, Qiao XJ, Kim HJ, et al. pparent diffusion coefficient histogram analysis stratifies progression-free and overall survival in patients with recurrent GM treated with bevacizumab: a multicenter study. J Neurooncol 2012; 108: Qayyum, oakley FV, Lu Y, et al. Organ-confined prostate cancer: effect of prior transrectal biopsy on endorectal MRI and MR spectroscopic imaging. JR 2004; 183: Turkbey, Shah VP, Pang Y, et al. Is apparent diffusion coefficient associated with clinical risk scores for prostate cancers that are visible on 3-T MR images? Radiology 2011; 258: Tamada T, Sone T, Jo Y, et al. pparent diffusion coefficient values in peripheral and transition zones of the prostate: comparison between normal and malignant prostatic tissues and correlation with histologic grade. J Magn Reson Imaging 2008; 28: Zelhof, Pickles M, Liney G, et al. orrelation of diffusion-weighted magnetic resonance data with cellularity in prostate cancer. JU Int 2009; 103: JR:203, July 2014