Investigation of prostate cancer using diffusion-weighted intravoxel incoherent motion imaging

Investigation of prostate cancer using diffusion-weighted intravoxel incoherent motion imaging

Available online at www.sciencedirect.com Magnetic Resonance Imaging 29 (2011) 1053 – 1058 Investigation of prostate cancer using diffusion-weighted...

624KB Sizes 2 Downloads 50 Views

Available online at www.sciencedirect.com

Magnetic Resonance Imaging 29 (2011) 1053 – 1058

Investigation of prostate cancer using diffusion-weighted intravoxel incoherent motion imaging Jörg Döpfert a,b,⁎, Andreas Lemke a , Anja Weidner c , Lothar R. Schad a a

Department of Computer Assisted Clinical Medicine, Heidelberg University, 68167 Mannheim, Germany ERC Project BiosensorImaging, Leibniz-Institut für Molekulare Pharmakologie, 13125 Berlin, Germany c Department of Clinical Radiology and Nuclear Medicine, University Medical Center Mannheim, Medical Faculty Mannheim, Heidelberg University, 68167 Mannheim, Germany Received 25 December 2010; revised 14 April 2011; accepted 1 June 2011 b

Abstract Purpose: The objective of this work was to evaluate the diagnostic performance of the intravoxel incoherent motion (IVIM) model to differentiate between healthy and malignant prostate tissue. Materials and Methods: Regions of interest were drawn in healthy and cancerous tissue of 13 patients with histologically proven prostate carcinoma and fitted to a monoexponential model [yielding the apparent diffusion coefficient (ADC)] and the IVIM signal equation (yielding the perfusion fraction f, the diffusion constant D and the pseudodiffusion coefficient of perfusion D⁎). Parameter maps were calculated for all parameters. Results: The ADC, D and f were significantly (Pb.005) lowered in cancerous tissue (1.01±0.22 μm2/ms, 0.84±0.19 μm2/ms and 14.27±7.10%, respectively) compared to benign tissue (1.49±0.17 μm2/ms, 1.21±0.22 μm2/ms and 21.25±8.32%, respectively). Parameter maps of D and f allowed for a delineation of the tumor, but showed higher variations compared to the ADC map. Conclusion: Apparent diffusion coefficient maps provide better diagnostic performance than IVIM maps for tumor detection. However, the results suggest that the reduction of the ADC in prostate cancer stems not only from changes in cellularity but also from perfusion effects. IVIM imaging might hold promise for the diagnosis of other prostatic lesions. © 2011 Elsevier Inc. All rights reserved. Keywords: Diffusion; MRI; Prostate cancer; Biexponential decay; IVIM, perfusion

1. Introduction Prostate cancer (PCa) is the fifth most common cancer worldwide and causes 6% of cancer deaths in men [1]. Transrectal ultrasonography (TRUS)-guided prostate biopsy [2] is accepted as the gold standard for diagnosis; however, the tumor detection rate remains unsatisfactory [3], and the procedure may furthermore be unpleasant to the patient. As an alternative, magnetic resonance imaging (MRI) is showing promise for PCa localization. Generally, T2weighted anatomical images of the gland are used to detect

⁎ Corresponding author. Leibniz-Institut für Molekulare Pharmakologie, Robert-Roessle-Str. 10, 13125 Berlin, Germany. E-mail address: [email protected] (J. Döpfert). 0730-725X/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.mri.2011.06.001

and stage PCa. However, this method lacks specificity (b27%) in small tumors (b1 cm in diameter) [4] and cannot distinguish between PCa and other lesions like prostatitis [5] and benign prostatic hyperplasia (BPH) [6]. To overcome these drawbacks, several other MR methods have been proposed [7–10], among them diffusion-weighted imaging (DWI). The apparent diffusion coefficient (ADC) — a parameter derived from diffusion-weighted images assuming a monoexponential model — reflects the incoherent motion of tissue water at the micron scale and is hence sensitive to pathophysiological changes on the cellular level. As reported by several groups, the additional use of the ADC (which is significantly decreased in PCa compared to healthy tissue) allows for an improvement of the tumor detection rate in comparison to T2-weighted imaging alone [11,12]. However, when using strong diffusion weightings (b values up to 3000 s/mm2), the signal decay curve was shown to be better

1054

J. Döpfert et al. / Magnetic Resonance Imaging 29 (2011) 1053–1058

described by a biexponential fit both in the healthy prostate [13] and PCa [14], suggesting the existence of a slow and a fast diffusing water pool owing to cellular restrictions. The intravoxel incoherent motion (IVIM) model predicts a third, much faster diffusing exponential component in the signal equation due to perfusion effects, which affects the overall signal predominantly at low b values [15]. In several abdominal organs like the liver and the pancreas, this perfusion component was recently observed and successfully described by the IVIM theory [16,17]. The additional information due to separation of “pure” molecular diffusion and perfusion effects allowed for a better characterization of lesions than the ADC. The purpose of this study was to determine whether IVIM imaging can play a role in differentiating between healthy and malignant prostatic tissue.

the MRI signal intensity as their motion due to molecular diffusion. Namely, if - the capillaries are pseudorandomly oriented - the blood flow changes capillary segments and hence its direction several times during the diffusion time Δ - the number of capillaries in a voxel is sufficiently large to allow for a statistical analysis then the movement of water in the capillary network mimics a diffusion process and can therefore be described by an exponentially decaying term in the signal equation. Combining this conclusion with Eq. (1) yields the threeparametric biexponential IVIM equation S = ð 1− f Þ  expð − bDÞ + f  expð − b  ð D∗ + DÞÞ; |fflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} S0 diffusion term

2. Theory

ð3Þ

perfusion term

The term “diffusion” denotes the phenomenon of randomly oriented translation of particles caused by thermal energy (Brownian molecular motion). The self-diffusion of water molecules in the presence of magnetic field gradients leads to a signal loss in MRI measurements that in the case of free diffusion (isotropic, unrestricted diffusion) can be described by a monoexponential function [18]

where f is the perfusion fraction, D the (molecular) diffusion coefficient and D⁎ the pseudodiffusion coefficient, which depends on the mean blood velocity v̄ and the mean capillary segment length l ̄ [15]. Since v̄ is considerably faster than the mean molecular diffusion velocity of water, the flow-related pseudodiffusion coefficient D⁎ is expected to be orders of magnitude greater than the tissue diffusion coefficient D. As a consequence, the second term (the perfusion-related component) in Eq. (3) becomes very small for high b values, and hence, perfusion effects are detectable at low b values.

S = expð − bDÞ: S0

3. Materials and methods

2.1. Basics of DWI

ð1Þ

Here, S is the measured signal intensity, S0 is the signal intensity without the influence of diffusion, D is the diffusion coefficient of water and the sequence-dependent b value characterizes the diffusion weighting. For the spin echo sequence with two rectangular diffusion gradients employed in this study, the b value can be calculated as follows [19] b = g2 G2 d2 ðΔ − d=3Þ;

ð2Þ

with γ denoting the gyromagnetic ratio, G the diffusion gradient amplitude, δ the diffusion gradient duration and Δ the time between leading edges of the diffusion gradient pulses, henceforth referred to as diffusion time. If the condition of free diffusion is violated (e.g., due to restrictions, exchange between compartments or perfusion), the value for D obtained by Eq. (1) no longer reflects “pure” diffusion properties. Thus, the parameter derived from Eq. (1) in biological tissue is usually called “apparent diffusion coefficient.” 2.2. The IVIM model The IVIM model, introduced by Le Bihan et al [15] in 1988, is based on the fact that, under certain assumptions, the motion of water molecules due to microcirculation of blood in the capillary network (perfusion) has a similar impact on

3.1. Patients In this study, clinical and imaging data of 13 patients (age 59–75 years; mean age 67 years) with biopsy-proven PCa were retrospectively evaluated, and informed consent was obtained from all patients. Besides T2-weighted MRI and DWI, all patients underwent prostatectomy (within 24 h after MR examination). All tumors were located in the peripheral zone (PZ). The mean prostate-specific antigen (PSA) level in these 13 patients was 10.4 ng/ml (range 2.8–21.5 ng/ml), the mean postoperative Gleason score was 7 (range 6–8) and the average number of prostate tumors was 2.7 (range 1–6). 3.2. MRI The MRI studies were carried out on a 3.0-T Scanner (Magnetom Tim Trio, Siemens Medical Solutions, Erlangen, Germany) using a standard six-channel pelvic phased array coil in conjunction with an endorectal coil (MedRad Inc., Pittsburgh, PA, USA). The endorectal coil was inserted after digital rectal examination and was inflated with 40–80 ml of air. To suppress bowel movement, up to 40 mg butylscopolammonium bromide (Buscopan, Boehringer Ingelheim GmbH, Ingelheim, Germany) was given intravenously if the patient had no contraindication. All patients underwent a routine prostatic MR protocol, including conventional

J. Döpfert et al. / Magnetic Resonance Imaging 29 (2011) 1053–1058 Table 1 Mean, standard deviation and range of the fitting parameters for the 13 patients ROI in PCa 2

ROI in healthy tissue

1.49±0.17 [1.19–1.71] ADC (μm /ms) 1.01±0.22 [0.61–1.41] D (μm2/ms) 0.84±0.19 [0.46–1.14] 1.21±0.22 [0.87–1.56] D* (μm2/ms) 7.52±4.77 [3.00–15.62] 6.82±2.78 [3.00–11.99] f (%) 14.27±7.10 [3.73–25.53] 21.25±8.32 [0.01–32.58]

P .0007 .0024 .7910 .0024

ROIs were drawn in PCa and contralateral healthy tissue. The ADC and the IVIM parameters D and f are significantly decreased in PCa compared to healthy tissue (Pb.05).

T2-weighted anatomical images in three orthogonal planes (TR=5170 ms, TE=101 ms, FOV=200×200 mm2, matrix size=320×320, slice thickness/gap=3/0.6 mm, seven averages, bandwidth=200 Hz/pixel and an acquisition time of 11

1055

min for all three planes) and diffusion-weighted images transverse to the prostate (TR=2600 ms, TE=66 ms, FOV=204×204 mm2 matrix size=136×136, slice thickness/ gap=3/0 mm, eight averages, bandwidth=1415 Hz/pixel and an acquisition time of 4.5 min) with b values of 0, 50, 500 and 800 s/mm2. Diffusion weighting was accomplished using a Stejskal–Tanner spin echo diffusion preparation with two monopolar diffusion gradient pulses, followed by a singleshot echo-planar imaging readout. For each b value, diffusionweighted images were acquired with three orthogonal gradient directions resulting in rotationally invariant trace images. 3.3. Data analysis Regions of PCa were delineated by an experienced pathologist using standard step section histopathology. On

A

B

C

D

E

F

Fig. 1. Prostate of an exemplary patient. (A) T2-weighted anatomical image. The tumor is marked by a red arrow. (B) b=0 s/mm2 image. (C) ADC map with ROIs drawn in cancerous (red) and healthy tissue (green). The tumor corresponds to the hypointense area. (D) D map. The “coefficient of molecular diffusion” is decreased in cancerous tissue (arrow). (E) f map. The perfusion fraction is decreased in cancerous tissue (arrow). (F) D⁎ map. The “pseudodiffusion coefficient” exhibits large variances in the pixel intensities; structures are hardly identifiable.

1056

J. Döpfert et al. / Magnetic Resonance Imaging 29 (2011) 1053–1058

the basis of these findings, the lesions were identified on the T2-weighted images and ADC maps by a radiologist, and for each patient, one region of interest (ROI) was placed in the largest tumor found in the PZ and one in contralateral healthy tissue, leading to a total number of 13 data points for tumor and healthy tissue, respectively. The average ROI size was 95±36 mm2 in PCa and 87±38 mm2 in healthy tissue. Mean signal intensities over the ROIs were calculated for each b value on the diffusion-weighted images. By means of a nonlinear least squares fit, these data were fitted to Eq. (1) to extract the ADC and to Eq. (3) to extract D, D⁎ and f using the following lower and upper bounds for the parameters: D (0.1–100 μm2/ms), D⁎ (3–100 μm2/ms) and f (0%–100%). Additionally, parameter maps were obtained by fitting the DWI data pixel by pixel. Postprocessing and data analysis were performed using homemade software written in Matlab (Mathworks Inc., Natick, MA, USA).

anatomical image of the prostate in the traverse plane, and Fig. 1 (B) shows the image of the DWI sequence of the corresponding slice at b=0 s/mm2. The tumor is marked with an arrow. In Fig. 1 (C), the ADC map of the same slice is displayed. PCa is clearly identifiable as a hypointense region. Two ROIs are placed: one around the cancerous tissue (red) and one in contralateral healthy tissue (green). The IVIM fits of the two ROIs are shown in Fig. 2. In agreement with Table 1, f and D are decreased in PCa compared to contralateral benign tissue in this patient as well. The IVIM parameter maps of the same slice are presented in Fig. 1 (D)–(F). Notably, fits converged for all pixels in the region of the gland. In both the D and the f map [Fig. 1 (D) and (E), respectively], PCa is identifiable as hypointense region. However, the D⁎ map shown in Fig. 1 (F) exhibits large variances leading to a poor image quality; the structure of the prostate is hardly identifiable.

3.4. Statistical analysis Statistical analysis of the data was performed using MedCalc (MedCalc Software bvba, Mariakerke, Belgium). A nonparametric pairwise Wilcoxon rank–sum test was used to detect whether there were significant differences between PCa and contralateral healthy tissue in the parameters ADC, D, D⁎ and f. A P value of less or equal .05 was considered statistically significant. 4. Results Table 1 summarizes the results obtained from ROIs in PCa and contralateral healthy tissue in all 13 patients. Along with the ADC, the IVIM parameters f and D turned out to be significantly decreased in PCa compared to healthy tissue, whereas no significant difference could be found for D⁎. In Fig. 1, representative maps for one single patient are presented as an example. Fig. 1 (A) shows a T2-weighted

Fig. 2. IVIM-fit according to Eq. (3) of the ROIs from Fig. 1 in healthy (green) and cancerous tissue (red).

5. Discussion In agreement with the findings of several other groups [20–22], the ADC determined in this study was significantly decreased in PCa compared to healthy prostate tissue. However, upon closer inspection, the signal decay measured with the diffusion-sensitized sequence turned out to be clearly nonmonoexponential at the b value range used. This observation confirms the results of the prostate study by Riches et al. [23], according to which the signal could be better described by two exponentially decaying terms rather than just by one. As a consequence, the ADC value measured strongly depends on the b value range chosen. This could explain the differences in the prostate ADCs reported in literature: For example, Kozlowski et al. [20] found the ADC to be 1.99±0.21 μm2/ms in the healthy PZ at b values of 0 and 600 s/mm2, whereas Gürses et al. [24] reported 1.61±0.35 μm2/ms at b values of 0 and 700 s/mm2. The ADC in the healthy prostate determined in our study agrees within the error limits with the value of Riches et al. (mean value between PZ and CZ: 1.69±0.49 μm2/ms), who also used b values between 0 and 800 s/mm2. Hence, it seems reasonable to compare prostate ADC values only when they are extracted from the same or at least similar b value range. This was already proposed by Zhang et al. [25] for the kidneys, who investigated the variability of the renal ADC at different b value distributions. However, since the ADC possibly varies with age [26] and glandular and stromal structure [27], also differences in patient populations may have contributed to the variability in the results of the different studies. In contrast, the IVIM model accounts for the nonmonoexponential behavior of the signal decay [28] and yielded encouraging results in recent studies concerning liver cirrhosis [17], assessment of ureteral obstruction [29] and pancreas lesions [16]. In this study, we could show that the IVIM parameters D and f were significantly lowered in PCa

J. Döpfert et al. / Magnetic Resonance Imaging 29 (2011) 1053–1058

compared to healthy prostate tissue. This result suggests that the aforementioned reduction of the ADC in PCa not only stems from changes in cellularity but also partly originates from perfusion effects. The values for D and f determined in this work are in agreement with the values presented by Riches et al. However, even though they observed a reduced perfusion fraction f in the tumor tissue, the only significant IVIM parameter they reported was the molecular diffusion coefficient D. This difference may be explained by the fact that they used PSA level and TRUS-guided biopsy to identify PCa, whereas in our study prostatectomy findings were considered, which enable a more accurate differentiation between the two groups. Another difference to their results is the value for D⁎, which deviates from the value determined in this work. This may originate from the fact that they chose far more b values in the regime sensitive to the perfusion component (0, 1, 2, 4, 10, 20, 50, 100 and 200 s/mm2), whereas in this work, just two b values (0 and 50 s/mm2) could be applied in this range. Nevertheless, the values for D⁎ spread widely in both of the studies. In addition to Riches et al., maps of the parameters D, f and D⁎, which are exemplarily shown in Fig. 1 (D), (E) and (F), respectively, could be calculated in this work. To our knowledge, these are the first IVIM parameter maps of the prostate published. In both the f and the D map, the carcinoma can be clearly delineated from surrounding healthy tissue. However, it is well known that a three-parametric fit is generally less stable than a one-parametric fit. Therefore, the IVIM maps exhibit larger variances in comparison to the ADC map. The fact that especially the parameter D⁎ is prone to errors [30], particularly when f in Eq. (3) gets very small, may explain the poor quality of the D⁎ map. In our opinion, the ADC map provides the best image quality regarding PCa detection among all generated parameter maps. Besides the aforementioned stability of the fit, this also may be due to the fact that the reduction of the ADC in PCa is composed of both the decrease in f and the decrease in D. Hence, the ADC map displays the changes in both of the parameters combined. It should be noted that this was not known a priori but is a result of the IVIM analysis. Nevertheless, the additional information provided by the IVIM model might be useful in identifying lesions other than PCa. For example, both T2-weithed images and ADC maps have only limited use for the discrimination between PCa and BPH [31]. Yet, a positron emission tomography study by Inaba [32] showed that both blood flow and blood volume differ significantly in PCa and BPH. Similarly, prostatitis can reduce the ADC and thus lead to a falsepositive appearance of PCa. As reported by Franiel et al. [8] by means of dynamic contrast-enhanced (DCE) MRI, both diseases showed a different perfusion, though. Hence, it seems promising to investigate the diagnostic performance of the IVIM model in differentiating between PCa, BPH and prostatitis in further studies, particularly because this

1057

technique is completely noninvasive and thus avoids the administration of contrast agents. In perfusion measurements using DCE MRI, PCa showed a higher blood volume compared to healthy prostate tissue [8]. This seems contradictory to our results and the trend observed by Riches et al. [23], which yielded a reduced perfusion fraction in PCa. However, Patel et al. [33] reported a discrepancy between IVIM parameters and perfusion parameters derived from DCE MRI, who directly compared the two methods in the liver. It has to be taken into account that f, just indicating the water fraction flowing through pseudorandomly oriented microcapillaries, does not directly correspond to any of the DCE parameters [33] and is dependent on the transversal relaxation time T2 of the compartments [28]. For all that, even if the IVIM model does not allow for a direct measurement of perfusion, the separation between fast and slowly diffusing water yields extra information about normal and pathological tissue that the ADC cannot provide. Limitations of our study include the fact that all images were acquired using a clinical standard protocol including only four fixed b values, from which just two (0 and 50 s/ mm 2 ) were in the range sensitive to the perfusion component. At first glance, this seems rather few to reliably determine f, D and D⁎. However, Zhang et al. [34] used Monte Carlo simulations based on parameters obtained in renal lesions to find an optimal b value distribution for the biexponential analysis of diffusionweighted images and showed that exactly four b values were best suited to estimate IVIM parameters. Furthermore, the pattern used in this study was similar to the optimal sampling pattern they reported (0, 51, 259 and 800 s/mm2). To verify these results in the case of prostatic lesions, a study should be performed comparing variances of IVIM fitting parameters obtained with different numbers of b values at identical scan times. Moreover, only a small number of patients (N=13) were involved in this study. A larger patient population with a broad range of tumor grades is needed to evaluate the usefulness of the IVIM model in the clinical setting.

6. Conclusion In conclusion, our results confirm that the signal decay observed within a b value range of [0–800 s/mm2] can be described well by the biexponential IVIM model, which allows for a completely noninvasive measurement of perfusion-related parameters. The significant reduction of the parameters D and f in PCa compared to healthy prostate tissue suggests that the well-known decrease of the ADC stems not only from changes in cellularity but also from perfusion effects. In both the D and the f map PCa is delineable, but best image quality regarding tumor identification is provided by the ADC map. Further study is required

1058

J. Döpfert et al. / Magnetic Resonance Imaging 29 (2011) 1053–1058

to evaluate the diagnostic performance of IVIM imaging in BPH and prostatitis detection. References [1] Ferlay J, Shin HR, Bray F, Forman D, Mathers C, Parkin DM. Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008. Int J Cancer 2010;127(12):2893–917. [2] Hodge KK, McNeal JE, Terris MK, Stamey TA. Random systematic versus directed ultrasound guided transrectal core biopsies of the prostate. J Urol 1989;142(1):71–4. [3] Eskicorapci SY, Baydar DE, Akbal C, Sofikerim M, Günay M, Ekici S, et al. An extended 10-core transrectal ultrasonography guided prostate biopsy protocol improves the detection of prostate cancer. Eur Urol 2004;45(4):444–9. [4] Nakashima J, Tanimoto A, Imai Y, Mukai M, Horiguchi Y, Nakagawa K, et al. Endorectal MRI for prediction of tumor site, tumor size, and local extension of prostate cancer. Urology 2004;64(1):101–5. [5] Casciani E, Polettini E, Bertini L, Pansadoro V, Gualdi GF. Granulomatous prostatitis: a pitfall in endorectal MR imaging and 3D MR spectroscopic imaging. Eur J Radiol Extra 2005;54(3):111–4. [6] Claus FG, Hricak H, Hattery RR. Pretreatment evaluation of prostate cancer: role of MR imaging and 1H MR spectroscopy. Radiographics 2004;24(1):167–80. [7] Kim CKMD, Park BKMD, Kim BMD. Localization of prostate cancer using 3T MRI: comparison of T2-weighted and dynamic contrastenhanced imaging. J Comput Assist Tomogr 2006;30(1):7–11. [8] Franiel T, Ludemann L, Rudolph B, Rehbein H, Staack A, Taupitz M, et al. Evaluation of normal prostate tissue, chronic prostatitis, and prostate cancer by quantitative perfusion analysis using a dynamic contrast-enhanced inversion-prepared dual-contrast gradient echo sequence. Invest Radiol 2008;43(7):481–7. [9] Carroll PR, Coakley FV, Kurhanewicz J. Magnetic resonance imaging and spectroscopy of prostate cancer. Rev Urol 2006;8(1):4–10. [10] Swanson MG, Vigneron DB, Tabatabai ZL, Males RG, Schmitt L, Carroll PR, et al. Proton HR-MAS spectroscopy and quantitative pathologic analysis of MRI/3D-MRSI-targeted postsurgical prostate tissues. Magn Reson Med 2003;50(5):944–54. [11] Morgan VA, Kyriazi S, Ashley SE, DeSouza NM. Evaluation of the potential of diffusion-weighted imaging in prostate cancer detection. Acta Radiol 2007;48(6):695–703. [12] Shimofusa RMD, Fujimoto HMD, Akamata HMD, Motoori KMD, Yamamoto SMD, Ueda TMD, et al. Diffusion-weighted imaging of prostate cancer. J Comput Assist Tomogr 2005;29(2):149–53. [13] Mulkern RV, Barnes AS, Haker SJ, Hung YP, Rybicki FJ, Maier SE, et al. Biexponential characterization of prostate tissue water diffusion decay curves over an extended b-factor range. Magn Reson Imaging 2006;24(5):563–8. [14] Shinmoto H, Oshio K, Tanimoto A, Higuchi N, Okuda S, Kuribayashi S, et al. Biexponential apparent diffusion coefficients in prostate cancer. Magn Reson Imaging 2009;27(3):355–9. [15] Le Bihan D, Breton E, Lallemand D, Aubin ML, Vignaud J, LavalJeantet M. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology 1988;168(2):497–505. [16] Lemke A, Laun FB, Klau M, Re TJ, Simon D, Delorme S, et al. Differentiation of pancreas carcinoma from healthy pancreatic tissue using multiple b-values: comparison of apparent diffusion coefficient and intravoxel incoherent motion derived parameters. Invest Radiol 2009;44(12):769–75.

[17] Luciani A, Vignaud A, Cavet M, Nhieu JT, Mallat A, Ruel L, et al. Liver cirrhosis: intravoxel incoherent motion MR imaging — pilot study. Radiology 2008;249(3):891–9. [18] Le Bihan D. Diffusion and perfusion magnetic resonance imaging: applications to functional MRI. New York: Raven Press; 1999. [19] Stejskal EO, Tanner JE. Spin diffusion measurements: spin echoes in the presence of a time-dependent field gradient. J Chem Physics 1965; 42(1):288–92. [20] Kozlowski P, Chang SD, Goldenberg SL. Diffusion-weighted MRI in prostate cancer — comparison between single-shot fast spin echo and echo planar imaging sequences. Magn Reson Imaging 2008;26(1): 72–6. [21] Xu J, Humphrey PA, Kibel AS, Snyder AZ, Narra VR, Ackerman JJH, et al. Magnetic resonance diffusion characteristics of histologically defined prostate cancer in humans. Magn Reson Med 2009;61(4): 842–50. [22] Sato C, Naganawa S, Nakamura T, Kumada H, Miura S, Takizawa O, et al. Differentiation of noncancerous tissue and cancer lesions by apparent diffusion coefficient values in transition and peripheral zones of the prostate. J Magn Reson Imaging 2005;21(3):258–62. [23] Riches SF, Hawtin K, Charles-Edwards EM, de Souza NM. Diffusionweighted imaging of the prostate and rectal wall: comparison of biexponential and monoexponential modelled diffusion and associated perfusion coefficients. NMR Biomed 2009;22(3):318–25. [24] Gurses B, Kabakci N, Kovanlikaya A, Firat Z, Bayram A, Ulug AM, et al. Diffusion tensor imaging of the normal prostate at 3 Tesla. Eur Radiol 2008;18(4):716–21. [25] Zhang JL, Sigmund EE, Chandarana H, Rusinek H, Chen Q, Vivier PH, et al. Variability of renal apparent diffusion coefficients: limitations of the monoexponential model for diffusion quantification. Radiology 2010;254(3):783–92. [26] Tamada T, Sone T, Toshimitsu S, Imai S, Jo Y, Yoshida K, et al. Agerelated and zonal anatomical changes of apparent diffusion coefficient values in normal human prostatic tissues. J Magn Reson Imaging 2008; 27(3):552–6. [27] Ren J, Huan Y, Wang H, Zhao H, Ge Y, Chang Y, et al. Diffusionweighted imaging in normal prostate and differential diagnosis of prostate diseases. Abdom Imaging 2008;33(6):724–8. [28] Lemke A, Laun FB, Simon D, Stieltjes B, Schad LR. An in vivo verification of the intravoxel incoherent motion effect in diffusionweighted imaging of the abdomen. Magn Reson Med 2010;64(6): 1580–5. [29] Thoeny HC, Binser T, Roth B, Kessler TM, Vermathen P. Noninvasive assessment of acute ureteral obstruction with diffusion-weighted MR imaging: a prospective study. Radiology 2009;252(3):721–8. [30] King MD, van Bruggen N, Busza AL, Houseman J, Williams SR, Gadian DG. Perfusion and diffusion MR imaging. Magn Reson Med 1992;24(2):288–301. [31] Kim JK, Jang YJ, Cho G. Multidisciplinary functional MR imaging for prostate cancer. Korean J Radiol 2009;10(6):535–51. [32] Inaba T. Quantitative measurements of prostatic blood flow and blood volume by positron emission tomography. J Urol 1992;148(5): 1457–60. [33] Patel J, Sigmund EE, Rusinek H, Oei M, Babb JS, Taouli B. Diagnosis of cirrhosis with intravoxel incoherent motion diffusion MRI and dynamic contrast-enhanced MRI alone and in combination: preliminary experience. J Magn Reson Imaging 2010;31(3):589–600. [34] Zhang J, Sigmund E, Chandarana H, Rusinek H, Guo H, Storey P, et al. Optimization of the b-sampling for bi-exponential analysis of diffusion-weighted imaging. Int Soc Magn Reson Med 2010:4706.