CME JOURNAL OF MAGNETIC RESONANCE IMAGING 40:1375–1381 (2014)
Original Research
Apparent Diffusion Coefficient Reproducibility of the Pancreas Measured at Different MR Scanners Using Diffusion-Weighted Imaging Xiao-Hua Ye, MD,1 Jia-Yin Gao, MD,1 Zheng-Han Yang, MD, PhD,1* and Yuan Liu, MD2 Purpose: To evaluate the reproducibility of the pancreatic apparent diffusion coefficient (ADC) measured at different MR scanners. Materials and Methods: Twenty-four healthy volunteers underwent three consecutive diffusion-weighted imaging (DWI) at a GE 1.5 Tesla (T), a Siemens 1.5T and a Philips 3.0T (session 1), and imaged again using the same protocol at the same GE 1.5T (session 2) 12 days later. The ADC values of pancreas were measured at all three MR scanners. Paired-sample t-test and the Bland-Altman method were used for ADC data analysis. Results: The individual mean ADC values of pancreatic head, body, and tail (in 103mm2/s) measured at GE 1.5T (2.24, 2.01, 1.88 for observer 1 and 2.23, 2.00, 1.92 for observer 2) and Siemens 1.5T (2.24, 2.04, 1.84 for observer 1 and 2.20, 1.98, 1.84 for observer 2) were significantly higher than those at Philips 3.0T (2.06, 1.80, 1.56 for observer 1 and 2.02, 1.79, 1.60 for observer 2) (P ¼ 0.000–0.008). There was no significant difference of ADC values either between GE 1.5T and Siemens 1.5T (P ¼ 0.115–0.966), or between imaging session 1 and 2 at GE 1.5T (P ¼ 0.072–0.938). The range of mean difference 6 limits of agreement (in 103mm2/s) was 0.07–0.04 6 0.39–0.53 between two 1.5T scanners, and 0.04–0.04 6 0.24–0.47 between two imaging sessions at GE 1.5T. Conclusion: The measured ADC values of pancreas are affected by the field strength of scanner, but show good reproducibility between different MR systems with same field strength and at the same MR system over time. Key Words: pancreas; diffusion-weighted imaging; apparent diffusion coefficient J. Magn. Reson. Imaging 2014;40:1375–1381. C 2013 Wiley Periodicals, Inc. V
1 Department of Radiology, Beijing Hospital and the 5th Clinical School of Peking University, Beijing, China. 2 Department of Ultrasound, Beijing Hospital and the 5th Clinical School of Peking University, Beijing, China. Contract grant sponsor: The Nature and Science Foundation of Beijing; Contract grant number: 7102138. *Address reprint requests to: Z.-H.Y., No. 1 Dahua Road, Dongdan, Beijing, China 100730. E-mail: [email protected] Received April 19, 2013; Accepted October 7, 2013. DOI 10.1002/jmri.24492 View this article online at wileyonlinelibrary.com. C 2013 Wiley Periodicals, Inc. V
DIFFUSION-WEIGHTED IMAGING (DWI) is used to assess the random motion of water molecules in the body with quantitative measurement of apparent diffusion coefficient (ADC) values (1,2). DWI, which derives its image contrast from the difference in the motion of water molecules between tissues, can be performed quickly and does not require the administration of exogenous contrast medium. The technique yields qualitative and quantitative information that reflects changes at cellular level, and provides insights about tumor cellularity and the integrity of cell membranes (3,4). DWI has been widely used in the central nervous system, particularly in the diagnosis of acute ischemic stroke (5–7). However, DWI of the upper abdomen has been a technical challenge due to respiration, bowel peristalsis, blood flow and long acquisition time (8). With the development of faster techniques and sequences in MR systems, DWI of the upper abdomen has become feasible in recent years (8–10). The application of DWI in the pancreatic diseases has been reported in several studies. Measurement of ADC has been reported to be useful in detecting pancreatic carcinoma (11,12), in characterizing solid and cystic lesions of the pancreas (13–21), in assessing pancreatic exocrine function in patients with chronic pancreatitis (22,23), and in predicting and monitoring therapeutic efficacy of pancreatic diseases (19,24). Diffusion-weighted MR imaging could be used as a quantitative tool for differential diagnosis and followup of pancreatic diseases. However, the ADC measurement may be affected by many potential and confounding factors including field strength and DWI technique (b-value, diffusion encoding direction, breathhold or free-breathing or respiratory triggered acquisition, parallel imaging acceleration, etc.) (25,26). Additionally, the magnetic field inhomogeneity, image noise and artifacts may also influence the measured ADC value (4,25,26). These factors may potentially alter the measured ADC value even in the absence of change in diffusion. If quantitative abdominal DWI sought to be clinically useful, particularly for predicting and monitoring therapeutic effects, understanding the impact of various factors to the measured ADC value is very important. Although some studies have shown good reproducibility
1375
1376
Ye et al.
Table 1 Parameters of Three MR Scanners* Parameter Gradient strength (mT/m) Gradient slew rate (mT/m.ms) DWI sequence Diffusion time parameters (ms) Shortest TE (ms) Diffusion gradient directions B-value (s/mm2) Repetition time (TR) Section thickness (mm) Intersection gap (mm) Field of view (cm) Matrix No. of signal averages Parallel acceleration factor Scanning time (s)
GE1.5T
SI1.5T
PH3.0T
23/40 80/150 Respiratory triggered SS-SE-EPI d ¼ 32.2, D ¼ 38.6
30 100 3-scan trace twice-refocused spin echo SS-SE-EPI (31) d1 ¼ 6, d2 ¼ 14, d3 ¼ 16, d4 ¼ 4.5 77 S/I, R/L, A/P 0 and 600 2 respiratory cycles 6 1.5 38 38 128 128 4 2 120
30/60 200/100 Respiratory triggered SS-SE-EPI d ¼ 15.6, D ¼ 26
60.8 S/I, R/L, A/P 0 and 600 2 respiratory cycles 6 1.5 38 38 128 128 4 2 96
48 S/I, R/L, A/P 0 and 600 2 respiratory cycles 6 1.5 38 38 128 128 4 2 102
*SI1.5T ¼ Siemens 1.5T; PH3.0T ¼ Philips 3.0T; d ¼ gradient time; D ¼ gradient interval; (31) ¼ reference (31).
of ADC measurements using DWI in certain abdominal organs (26–29), little is known about the reproducibility of ADC measurements for the pancreas. DWI is often performed at different MR systems in clinical practice, and better understanding the reproducibility of ADC measurements is necessary to accurately interpret DWI. Therefore, the purpose of our study was to assess the reproducibility of ADC measurements at different MR systems for the pancreas. MATERIALS AND METHODS MR Scanning This prospective study was approved by the local institutional review board. Written informed consent was obtained from all volunteers. During September and December of 2009 (session 1), 24 healthy volunteers (15 male, 9 female; mean age, 28 years; age range, 24–36 years) were included in this study. All volunteers fasted for 4–6 h before scanning, and underwent fat suppressed T1WI and T2WI scans to exclude any pancreatic abnormalities. Each subject then underwent three consecutive DWI at a GE 1.5 Tesla (T) (Signa Twin-Speed HD, GE Healthcare, Milwaukee, WI), a Siemens 1.5T (Magnetom Espree, Siemens Healthcare, Erlangen, Germany) and a Philips 3.0T (Achieva Dual, Philips Healthcare, Best, The Netherlands) magnetic resonance system. An eightelement phased array coil was used for signal reception. The parameters for each MR scanner are shown in Table 1. The same spectrally selective fat saturation technique was used at all three MR scanners. The ADC values for each DWI series were automatically calculated by the MR system and displayed as corresponding ADC maps (Fig. 1). The same formula for the calculation of ADC [SI ¼ SI0exp(-b ADC)] was used at all three MR systems, and the same threshold setting was used to avoid pixels with too low SNR when calculating the ADC.
After 12 6 2 (mean 6 standard deviation) days, all volunteers were imaged again (session 2) with the earlier described protocols at the same GE 1.5T MR system and the same software. As before, the ADC values for each DWI series were calculated automatically by the MR system and displayed as corresponding ADC maps. Data Analysis Quantitative analysis was performed at freestanding MR workstations by two radiology residents (J.Y.G., Y.L., with 4 and 2 years of experience in abdominal imaging, respectively), who were blinded to the subject identification information and imaging session. Region-of-interest (ROI) measurements were obtained in the pancreatic head, body and tail. Three nonoverlapping circular ROIs with a standardized size of 10 pixels were placed in homogeneous artifact-free areas, with large blood vessels excluded, on the images acquired without diffusion weighted gradient (b ¼ 0). By applying the copy and paste function of the workstation, ROIs were transferred to the same locations on the corresponding ADC maps (Fig. 2). The resulting mean ADC values for each of the three ROIs in the three corresponding anatomic locations represented the ADC values of pancreatic head, body and tail in a single individual. The ROIs were placed as consistent as possible for different scans. Statistical Analysis Statistical analysis was performed using SPSS 17.0 (SPSS Inc., Chicago, IL) and MedCalc Software (MedCalc, Mariakerke, Belgium). First, we assessed the agreement between observer 1 and 2 of ADC values measured at all three MR scanners. Second, we compared the ADC values between MR scanners measured by the same observer during imaging session 1.
ADC Reproducibility of the Pancreas Using DWI
1377
Figure 1. Representative images acquired from a volunteer. Diffusion-weighted images (b ¼ 600s/mm2) (a–c) and corresponding ADC maps (d–f) obtained at GE 1.5T, Siemens 1.5T, and Philips 3.0T.
Third, the ADC data measured at different sessions (only at GE 1.5T) by the same observer were also compared. Intraclass correlation coefficient (ICC) was used to assess the interobserver agreement. Pairedsample t-test was used for comparison of ADC values between scanners and imaging sessions. A P value of less than 0.05 was considered to be statistically significant. The significance level for comparison between scanners was 0.017(0.05/3) after Bonferroni correction. Furthermore, reproducibility of ADC measurements was determined as mean absolute difference (bias) and 95% confidence interval of the mean difference (limits of agreement) according to the Bland-Altman method (25). RESULTS The ADC values of pancreas showed good correlation between observer 1 and 2 at all three MR scanners (Table 2). During imaging session 1, the ADC values of pancreatic head, body and tail measured by the same observer at GE 1.5T were significantly higher than those at Philips 3.0T (P ¼ 0.008, 0.000, 0.000 for observer 1 and 0.008, 0.001, 0.000 for observer 2).
The ADC values of pancreatic head, body and tail measured at Siemens 1.5T were also significantly higher than those at Philips 3.0T (P ¼ 0.006, 0.000, 0.000 for observer 1 and 0.006, 0.002, 0.000 for observer 2). There were no significant differences in the measured ADC values between GE 1.5T and Siemens 1.5T (P ¼ 0.966, 0.394, 0.448 for observer 1 and 0.597, 0.691, 0.115 for observer 2) (Table 3). Figure 3 showed the results of Bland-Altman reproducibility analysis of ADC measurements between two scanners with 1.5T. The range of mean difference 6 limits of agreement (in 103mm2/s) between two scanners with 1.5T were 0.04–0.04 6 0.39–0.51 (observer 1), and 0.07– 0.02 6 0.43–0.53 (observer 2). There were no significant differences in the measured ADC values by the same observer between imaging session 1 and 2 at GE 1.5T (P ¼ 0.938, 0.464, 0.431 for observer 1 and 0.727, 0.151, 0.072 for observer 2) (Table 4). Figure 4 showed the results of Bland-Altman reproducibility analysis of ADC measurements between two imaging sessions. The range of mean difference 6 limits of agreement (in 103mm2/s) between two imaging sessions at GE1.5T were 0.03– 0.04 6 0.35–0.47 (observer 1), and 0.04–0.04 6 0.24–0.27 (observer 2).
Figure 2. Three ROIs (1–3) were placed in the pancreatic tail on the echo-planar image acquired without diffusion weighted gradient (b ¼ 0) (a), which were copied and pasted to the same locations on the corresponding ADC map (b).
1378
Ye et al.
Table 2 ADC Values of Pancreas Measured by Different Observers at Three MR Scanners* Scanner Location
GE1.5 head
body
SI1.5T tail
head
body
PH3.0 tail
head
body
tail
Observer1 2.24 6 0.33 2.01 6 0.23 1.88 6 0.26 2.24 6 0.23 2.04 6 0.26 1.84 6 0.28 2.06 6 0.23 1.80 6 0.23 1.56 6 0.25 Observer2 2.23 6 0.34 2.00 6 0.17 1.92 6 0.22 2.20 6 0.25 1.98 6 0.21 1.84 6 0.25 2.02 6 0.19 1.79 6 0.25 1.60 6 0.18 ICC 0.875 0.747 0.781 0.726 0.495 0.910 0.796 0.763 0.756 P value 0.000 0.000 0.000 0.000 0.006 0.000 0.000 0.000 0.000 *Data are mean values 6 standard deviation ( 103mm2/s). SI1.5T ¼ Siemens 1.5T; PH3.0T ¼ Philips 3.0T; head ¼ pancreatic head; body ¼ pancreatic body; tail ¼ pancreatic tail; ICC ¼ intraclass correlation coefficient.
DISCUSSION In clinical practice, the ADC values are often measured by different observers during follow-up examinations; hence, it might be helpful to know the interobserver difference of ADC measurements. Our results showed good agreement between observers in ADC measurements of pancreas, which indicate that ADC measurements have good reproducibility and can be performed by different observers. However, it is important to standardize the method when ADC is measured by different observers. In our study, there were significant differences in the measured ADC values of pancreas between GE 1.5T and Philips 3.0T, also between Siemens 1.5T and Philips 3.0T. The ADC values measured at both 1.5T scanners were significantly higher than those at 3.0T. In the previous studies (26,27), the ADC values of pancreas measured at 1.5T scanner were higher than at 3.0T but not significantly, perhaps related to the smaller n of these prior studies. However, the difference of pancreatic ADC between 1.5T and 3.0T in our study cannot be entirely attributed to difference in field strength, differences in echo time (TE) and vendor may be confounding factors. In our study, most parameters were kept the same across different MR scanners except TE. Because DWI is known to be error prone when signal-to-noise ratio (SNR) is poor, we adopted the shortest obtainable TE to ensure sufficient SNR for DW images. However, the shortest obtainable TE in SS-SE-EPI DWI sequence was different across scanners and it was the longest on Siemens 1.5T scanner, which may be due to the differences in pulse sequence design and gradient performance. Additionally, the TE setting in SS-SEEPI DWI sequence on GE 1.5T scanner cannot be
modified. Therefore, the chosen TE was not the same on different scanners in our study. In the previous studies (26,27), the MR systems of 1.5T and 3.0T came from same vendor. But, in our study, they are from different vendors. In biological tissue, diffusion may be restricted. The measured ADC can be affected by the diffusion time determined by DWI pulse sequence and imaging protocols as an indication of restricted diffusion (30). If the diffusion time is different in two DWI protocols, the measured ADC could be different due to restricted diffusion. In our study, it is difficult to determine the exact diffusion time for the twice-refocused SS-SE-EPI sequence on Siemens 1.5T scanner (31). Diffusion time was also different on the other two scanners (27.9 ms on GE 1.5T scanner, 20.8 ms on Philips 3.0T scanner). More research is needed to confirm the field strength effect without the influence of diffusion time. In our study, there were no significant differences in the measured ADC values of pancreas between GE 1.5T and Siemens 1.5T, and the Bland-Altman method indicated good reproducibility. The reproducibility of ADC measurements was good when measured at different scanners with same field strength. Because, in clinical practice, the follow-up examination for the same individual may be performed on a MR scanner with same field strength but from different vendors, it is necessary to know the ADC reproducibility measured at different scanners with same field strength. Our results indicate that ADC measurements have good reproducibility and the comparison of ADC values can be performed at different MR systems with the same field strength. In our study, there were no significant differences in the measured ADC values of pancreas between
Table 3 ADC Values of Pancreas Measured by the Same Observer at Three MR Scanners* GE1.5 PH3.0T Anatomic location 1
2
Pancreatic head Pancreatic body Pancreatic tail Pancreatic head Pancreatic body Pancreatic tail
GE1.5T 2.24 2.01 1.88 2.23 2.00 1.92
6 6 6 6 6 6
0.33 0.23 0.26 0.34 0.17 0.22
SI1.5T 2.24 2.04 1.84 2.20 1.98 1.84
6 6 6 6 6 6
0.23 0.26 0.28 0.25 0.21 0.25
PH3.0T 2.06 1.80 1.56 2.02 1.79 1.60
6 6 6 6 6 6
0.23 0.23 0.25 0.19 0.25 0.18
SI1.5T PH3.0T
GE1.5T SI1.5T
t value
P value
t value
P value
t value
P value
2.924 4.866 6.323 2.899 3.988 7.218
0.008 0.000 0.000 0.008 0.001 0.000
3.053 5.247 4.456 3.027 3.510 4.357
0.006 0.000 0.000 0.006 0.002 0.000
0.043 0.868 0.772 0.537 0.403 1.636
0.966 0.394 0.448 0.597 0.691 0.115
*Data are mean values 6 standard deviation ( 103mm2/s). SI1.5T ¼ Siemens 1.5T; PH3.0T ¼ Philips 3.0T; 1 ¼ observer 1; 2 ¼ observer 2.
ADC Reproducibility of the Pancreas Using DWI
1379
Figure 3. Reproducibility of pancreatic ADC measured at GE 1.5T and Siemens 1.5T scanners. Bland-Altman plots of difference of ADC measurements (y-axis) against mean ADC measurements (x-axis), with mean absolute difference (bias) (continuous line) and 95% confidence interval of the mean difference (limits of agreement) (dashed lines) for pancreatic head (a), pancreatic body (b), and pancreatic tail (c). Table 4 ADC Values of Pancreas Measured by the Same Observer in Different Time at GE1.5T* Anatomic location Pancreatic head Pancreatic body Pancreatic tail Pancreatic head 2 Pancreatic body Pancreatic tail 1
Session 1 2.24 2.01 1.88 2.23 2.00 1.92
6 6 6 6 6 6
0.33 0.23 0.26 0.34 0.17 0.22
Session 2 2.24 2.04 1.85 2.24 2.04 1.87
6 6 6 6 6 6
t value
0.32 0.078 0.25 0.744 0.22 0.801 0.34 0.354 0.15 1.484 0.21 1.889
P value 0.938 0.464 0.431 0.727 0.151 0.072
*Data are mean values 6 standard deviation ( 103mm2/s). 1 ¼ observer 1; 2 ¼ observer 2.
imaging session 1 and 2 at GE 1.5T, and the BlandAltman method indicated good reproducibility. The ADC measurements have good reproducibility measured at different sessions using the same MR scanner. The ADC values for each individuals were fairly stable over a period (up to 2 weeks), which was similar as shown in the previous studies (27,28). Rosenkrantz et al found that there was no significant difference in reproducibility of ADC for pancreas between 1.5T and 3.0T measured with a mean time interval of 12 days (27). Braithwaite et al found that there was no significant effect on ADCs of pancreas measured over a short-term (separated by five sequences within one imaging session) or
midterm (period between two imaging sessions with a mean time interval of 147 days) for a given individual (28). The ADC values of pancreatic head, body and tail measured in our study are different from those in the study by Dale et al (26). Dale et al measured the pancreatic ADC using a free-breathing technique at both 1.5T and 3.0T scanners, and indicated that the choice of b-values has a highly significant effect on ADC measurements with the general trend of decreasing measured ADC value with increasing b-value. With the b-value of 0 and 400, the mean ADC values of pancreatic head, body and tail were 2.65 103mm2/s, 2.89 103mm2/s, and 2.33 103mm2/s at 1.5T, 2.43 103mm2/s, 2.66 103mm2/s, and 2.24 103mm2/s at 3.0T. However, with the b-value of 0 and 800, the mean ADC values of pancreatic head, body and tail were 1.90 103mm2/s, 2.12 103mm2/s, and 1.75 103mm2/s at 1.5T, 1.71 103mm2/s, 1.79 103mm2/s, and 1.70 103mm2/s at 3.0T. The difference of pancreatic ADC between our study and Dale’s study may be due to the difference in b-values, and all results were mostly consistent with the effect of b-values on measured ADC value. Furthermore, the free-breathing technique adopted by Dale et al was substantially different from the respiratory-triggered technique in our study, which may also introduce some significant effects.
Figure 4. Reproducibility of pancreatic ADC measured at GE 1.5T with a time interval of 12 days. Bland-Altman plots of difference of ADC measurements (y-axis) against mean ADC measurements (x-axis), with mean absolute difference (bias) (continuous line) and 95% confidence interval of the mean difference (limits of agreement) (dashed lines) for pancreatic head (a), pancreatic body (b), and pancreatic tail (c).
1380
It is well known that more reliable calculations of ADC are possible if DWI sequences with more bvalues are used. However, we used a DWI sequence with only two b-values (0 and 600 s/mm2). This is because it is more typical in the current clinical practice. Additionally, only two b-values are available on our GE 1.5T scanner, so we used two b-values (0 and 600) for all three MR scanners to keep imaging parameters consistent. However, such choice for only two b-values may reduce confidence in the obtained ADC values, and the potential influence from the perfusion components could not be better characterized. Although abdominal DWI has been found in several studies to be robust and repeatable, our study indicates it is important to be careful in interpreting the change of measured ADC values. The difference in ADC values may be merely cause by the changes of MR scanner and/or measurement technique rather than the disease progression or treatment effect. There were several limitations in our study. First, the MR scanners were from different vendors in the study for assessing the effect of field strength on ADC measurement, so it was difficult to know whether the discrepancy of ADC values was due to one of the factors or a combination of both. Second, to keep the imaging parameters as consistent as possible, we performed all imaging examinations with the same scanning parameters except TE. This ideal scenario may not be achievable in actual clinical practice. Third, reproducibility beyond 2 weeks was not evaluated. The period of up to 2 weeks allows for the assessment of early response of treatment but not for long-term follow-up. Fourth, we did not evaluate the effect of different diffusion imaging techniques on the reproducibility of ADC measurement. It is known that the number of b values, the diffusion encoding direction, the breath modality, the use of parallel acquisition technique could affect the measured ADC value (4,25,26), so it is possible that they also affect reproducibility. Finally, our study population was relatively small and consisted of young, healthy subjects, agerelated effects could not be tested. Also what we really need to know is the reproducibility in clinical patients with disease, so further research is needed. In conclusion, the field strength of MR scanner may affect the measured ADC values of pancreas. The ADC values of pancreas show good reproducibility when measured at different MR systems with same field strength or over a period within 2 weeks at the same MR system. The specific results are only applicable to pancreatic ADC measurement using a similar DWI sequence with two b-values as in our study. REFERENCES 1. Lyng H, Haraldseth O, Rofstad EK. Measurement of cell density and necrotic fraction in human melanoma xenografts by diffusion weighted magnetic resonance imaging. Magn Reson Med 2000;43: 828–836. 2. Ichikawa T, Erturk SM, Motosugi U, et al. High-b-value diffusionweighted MRI in colorectal cancer. AJR Am J Roentgenol 2006; 187:181–184. 3. Koh DM, Takahara T, Imai Y, Collins DJ. Practical aspects of assessing tumors using clinical diffusion-weighted imaging in the body. Magn Reson Med Sci 2007;6:211–224.
Ye et al. 4. Koh DM, Collins DJ. Diffusion-weighted MRI in the body: applications and challenges in oncology. AJR Am J Roentgenol 2007; 188:1622–1635. 5. Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E, LavalJeantet M. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurological disorders. Radiology 1986;161:401–407. 6. Warach S, Chien D, Li W, Ronthal M, Edelman RR. Fast magnetic resonance diffusion-weighted imaging of acute human stroke. Neurology 1992;42:1717–1723. 7. Buckley BT, Wainwright A, Meagher T, Briley D. Audit of a policy of magnetic resonance imaging with diffusion-weighted imaging as first-line neuroimaging for in-patients withclinically suspected acute stroke. Clin Radiol 2003;58:234–237. 8. Chow LC, Bammer R, Moseley ME, Sommer FG. Single breathhold diffusion-weighted imaging of the abdomen. J Magn Reson Imaging 2003;18:377–382. 9. Ichikawa T, Haradome H, Hachiya J, Nitatori T, Araki T. Diffusion-weighted MR imaging single-shot echo-planar imaging in the upper abdomen: preliminary clinical experience in 61 patients. Abdom Imaging 1999;24:456–461. 10. Perman WH, Balci NC, Akduman I, Kuntz E. Magnetic resonance measurement of diffusion in the abdomen. Top Magn Reson Imaging 2009;20:99–104. 11. Shinya S, Sasaki T, Nakagawa Y, Guiquing Z, Yamamoto F, Yamashita Y. Usefulness of diffusion-weighted imaging (DWI) for the detection of pancreatic cancer: 4 cases reports. Hepatogastroenterology 2008;55:282–285. 12. Matsuki M, Inada Y, Nakai G, et al. Diffusion-weighed MR imaging of pancreatic carcinoma. Abdom Imaging 2007;32: 481–483. 13. Muraoka N, Uematsu H, Kimura H, et al. Apparent diffusion coefficient in pancreatic cancer: characterization and histopathological correlations. J Magn Reson Imaging 2008;27:1302–1308. 14. Momtahen AJ, Balci NC, Alkaade S, Akduman EI, Burton FR. Focal pancreatitis mimicking pancreatic mass: magnetic resonance imaging (MRI)/magnetic resonance cholangiopancreatography (MRCP) findings including diffusion-weighted MRI. Acta Radiol 2008;49:490–497. 15. Takeuchi M, Matsuzaki K, Kubo H, Nishitani H. High-b-value diffusion-weighted magnetic resonance imaging of pancreatic cancer and mass-forming chronic pancreatitis: preliminary results. Acta Radiol 2008;49:383–386. 16. Fattahi R, Balci NC, Perman WH, et al. Pancreatic diffusionweighted imaging (DWI): comparison between mass-forming focal pancreatitis (FP), pancreatic cancer (PC), and normal pancreas. J Magn Reson Imaging 2009;29:350–356. 17. Lee SS, Byun JH, Park BJ, et al. Quantitative analysis of diffusion-weighted magnetic resonance imaging of the pancreas: usefulness in characterizing solid pancreatic masses. J Magn Reson Imaging 2008;28:928–936. 18. Kartalis N, Lindholm TL, Aspelin P, Permert J, Albiin N. Diffusion-weighted magnetic resonance imaging of pancreas tumours. Eur Radiol 2009;19:1981–1990. 19. Kamisawa T, Takuma K, Anjiki H, et al. Differentiation of autoimmune pancreatitis from pancreatic cancer by diffusion-weighted MRI. Am J Gastroenterol 2010;105:1870–1875. 20. Inan N, Arslan A, Akansel G, Anik Y, Demirci A. Diffusionweighted imaging in the differential diagnosis of cystic lesions of the pancreas. AJR Am J Roentgenol 2008;191:1115–1121. 21. Fatima Z, Ichikawa T, Motosugi U, et al. Magnetic resonance diffusion-weighted imaging in the characterization of pancreatic mucinous cystic lesions. Clin Radiol 2011;66:108–111. 22. Erturk SM, Ichikawa T, Motosugi U, Sou H, Araki T. Diffusion weighted MR imaging in the evaluation of pancreatic exocrine function before and after secretin stimulation. Am J Gastroenterol 2006;101:133–136. 23. Akisik MF, Aisen AM, Sandrasegaran K, et al. Assessment of chronic pancreatitis: utility of diffusion-weighted MR imaging with secretin enhancement. Radiology 2009;250:103–109. 24. Kim H, Morgan DE, Buchsbaum DJ, et al. Early therapy evaluation of combined anti-death receptor 5 antibody and gemcitabine in orthotopic pancreatic tumor xenografts by diffusion-weighted magnetic resonance imaging. Cancer Res 2008;68:8369–8376. 25. Kwee TC, Takahara T, Koh DM, Nievelstein RA, Luijten PR. Comparison and reproducibility of ADC measurements in breathhold,
ADC Reproducibility of the Pancreas Using DWI respiratory triggered, and free-breathing diffusion-weighted MR imaging of the liver. J Magn Reson Imaging 2008;28:1141–1148. 26. Dale BM, Braithwaite AC, Boll DT, Merkle EM. Field strength and diffusion encoding technique affect the apparent diffusion coefficient measurements in diffusion-weighted imaging of the abdomen. Invest Radiol 2010;45:104–108. 27. Rosenkrantz AB, Oei M, Babb JS, Niver BE, Taouli B. Diffusionweighted imaging of the abdomen at 3.0 Tesla: image quality and apparent diffusion coefficient reproducibility compared with 1.5 Tesla. J Magn Reson Imaging 2011;33:128–135. 28. Braithwaite AC, Dale BM, Boll DT, Merkle EM. Short- and midterm reproducibility of apparent diffusion coefficient measure-
1381 ments at 3.0-T diffusion-weighted imaging of the abdomen. Radiology 2009;250:459–465. 29. Yoshikawa T, Kawamitsu H, Mitchell DG, et al. ADC measurement of abdominal organs and lesions using parallel imaging technique. AJR Am J Roentgenol 2006;187:1521– 1530. 30. Norris DG. The effects of microscopic tissue parameters on the diffusion weighted magnetic resonance imaging experiment. NMR Biomed 2001;14:77–93. 31. Reese TG, Heid O, Weisskoff RM, Wedeen VJ. Reduction of eddycurrent-induced distortion in diffusion MRI using a twicerefocused spin echo. Magn Reson Med 2003;49:177–182.
Original Research
Apparent Diffusion Coefficient Reproducibility of the Pancreas Measured at Different MR Scanners Using Diffusion-Weighted Imaging Xiao-Hua Ye, MD,1 Jia-Yin Gao, MD,1 Zheng-Han Yang, MD, PhD,1* and Yuan Liu, MD2 Purpose: To evaluate the reproducibility of the pancreatic apparent diffusion coefficient (ADC) measured at different MR scanners. Materials and Methods: Twenty-four healthy volunteers underwent three consecutive diffusion-weighted imaging (DWI) at a GE 1.5 Tesla (T), a Siemens 1.5T and a Philips 3.0T (session 1), and imaged again using the same protocol at the same GE 1.5T (session 2) 12 days later. The ADC values of pancreas were measured at all three MR scanners. Paired-sample t-test and the Bland-Altman method were used for ADC data analysis. Results: The individual mean ADC values of pancreatic head, body, and tail (in 103mm2/s) measured at GE 1.5T (2.24, 2.01, 1.88 for observer 1 and 2.23, 2.00, 1.92 for observer 2) and Siemens 1.5T (2.24, 2.04, 1.84 for observer 1 and 2.20, 1.98, 1.84 for observer 2) were significantly higher than those at Philips 3.0T (2.06, 1.80, 1.56 for observer 1 and 2.02, 1.79, 1.60 for observer 2) (P ¼ 0.000–0.008). There was no significant difference of ADC values either between GE 1.5T and Siemens 1.5T (P ¼ 0.115–0.966), or between imaging session 1 and 2 at GE 1.5T (P ¼ 0.072–0.938). The range of mean difference 6 limits of agreement (in 103mm2/s) was 0.07–0.04 6 0.39–0.53 between two 1.5T scanners, and 0.04–0.04 6 0.24–0.47 between two imaging sessions at GE 1.5T. Conclusion: The measured ADC values of pancreas are affected by the field strength of scanner, but show good reproducibility between different MR systems with same field strength and at the same MR system over time. Key Words: pancreas; diffusion-weighted imaging; apparent diffusion coefficient J. Magn. Reson. Imaging 2014;40:1375–1381. C 2013 Wiley Periodicals, Inc. V
1 Department of Radiology, Beijing Hospital and the 5th Clinical School of Peking University, Beijing, China. 2 Department of Ultrasound, Beijing Hospital and the 5th Clinical School of Peking University, Beijing, China. Contract grant sponsor: The Nature and Science Foundation of Beijing; Contract grant number: 7102138. *Address reprint requests to: Z.-H.Y., No. 1 Dahua Road, Dongdan, Beijing, China 100730. E-mail: [email protected] Received April 19, 2013; Accepted October 7, 2013. DOI 10.1002/jmri.24492 View this article online at wileyonlinelibrary.com. C 2013 Wiley Periodicals, Inc. V
DIFFUSION-WEIGHTED IMAGING (DWI) is used to assess the random motion of water molecules in the body with quantitative measurement of apparent diffusion coefficient (ADC) values (1,2). DWI, which derives its image contrast from the difference in the motion of water molecules between tissues, can be performed quickly and does not require the administration of exogenous contrast medium. The technique yields qualitative and quantitative information that reflects changes at cellular level, and provides insights about tumor cellularity and the integrity of cell membranes (3,4). DWI has been widely used in the central nervous system, particularly in the diagnosis of acute ischemic stroke (5–7). However, DWI of the upper abdomen has been a technical challenge due to respiration, bowel peristalsis, blood flow and long acquisition time (8). With the development of faster techniques and sequences in MR systems, DWI of the upper abdomen has become feasible in recent years (8–10). The application of DWI in the pancreatic diseases has been reported in several studies. Measurement of ADC has been reported to be useful in detecting pancreatic carcinoma (11,12), in characterizing solid and cystic lesions of the pancreas (13–21), in assessing pancreatic exocrine function in patients with chronic pancreatitis (22,23), and in predicting and monitoring therapeutic efficacy of pancreatic diseases (19,24). Diffusion-weighted MR imaging could be used as a quantitative tool for differential diagnosis and followup of pancreatic diseases. However, the ADC measurement may be affected by many potential and confounding factors including field strength and DWI technique (b-value, diffusion encoding direction, breathhold or free-breathing or respiratory triggered acquisition, parallel imaging acceleration, etc.) (25,26). Additionally, the magnetic field inhomogeneity, image noise and artifacts may also influence the measured ADC value (4,25,26). These factors may potentially alter the measured ADC value even in the absence of change in diffusion. If quantitative abdominal DWI sought to be clinically useful, particularly for predicting and monitoring therapeutic effects, understanding the impact of various factors to the measured ADC value is very important. Although some studies have shown good reproducibility
1375
1376
Ye et al.
Table 1 Parameters of Three MR Scanners* Parameter Gradient strength (mT/m) Gradient slew rate (mT/m.ms) DWI sequence Diffusion time parameters (ms) Shortest TE (ms) Diffusion gradient directions B-value (s/mm2) Repetition time (TR) Section thickness (mm) Intersection gap (mm) Field of view (cm) Matrix No. of signal averages Parallel acceleration factor Scanning time (s)
GE1.5T
SI1.5T
PH3.0T
23/40 80/150 Respiratory triggered SS-SE-EPI d ¼ 32.2, D ¼ 38.6
30 100 3-scan trace twice-refocused spin echo SS-SE-EPI (31) d1 ¼ 6, d2 ¼ 14, d3 ¼ 16, d4 ¼ 4.5 77 S/I, R/L, A/P 0 and 600 2 respiratory cycles 6 1.5 38 38 128 128 4 2 120
30/60 200/100 Respiratory triggered SS-SE-EPI d ¼ 15.6, D ¼ 26
60.8 S/I, R/L, A/P 0 and 600 2 respiratory cycles 6 1.5 38 38 128 128 4 2 96
48 S/I, R/L, A/P 0 and 600 2 respiratory cycles 6 1.5 38 38 128 128 4 2 102
*SI1.5T ¼ Siemens 1.5T; PH3.0T ¼ Philips 3.0T; d ¼ gradient time; D ¼ gradient interval; (31) ¼ reference (31).
of ADC measurements using DWI in certain abdominal organs (26–29), little is known about the reproducibility of ADC measurements for the pancreas. DWI is often performed at different MR systems in clinical practice, and better understanding the reproducibility of ADC measurements is necessary to accurately interpret DWI. Therefore, the purpose of our study was to assess the reproducibility of ADC measurements at different MR systems for the pancreas. MATERIALS AND METHODS MR Scanning This prospective study was approved by the local institutional review board. Written informed consent was obtained from all volunteers. During September and December of 2009 (session 1), 24 healthy volunteers (15 male, 9 female; mean age, 28 years; age range, 24–36 years) were included in this study. All volunteers fasted for 4–6 h before scanning, and underwent fat suppressed T1WI and T2WI scans to exclude any pancreatic abnormalities. Each subject then underwent three consecutive DWI at a GE 1.5 Tesla (T) (Signa Twin-Speed HD, GE Healthcare, Milwaukee, WI), a Siemens 1.5T (Magnetom Espree, Siemens Healthcare, Erlangen, Germany) and a Philips 3.0T (Achieva Dual, Philips Healthcare, Best, The Netherlands) magnetic resonance system. An eightelement phased array coil was used for signal reception. The parameters for each MR scanner are shown in Table 1. The same spectrally selective fat saturation technique was used at all three MR scanners. The ADC values for each DWI series were automatically calculated by the MR system and displayed as corresponding ADC maps (Fig. 1). The same formula for the calculation of ADC [SI ¼ SI0exp(-b ADC)] was used at all three MR systems, and the same threshold setting was used to avoid pixels with too low SNR when calculating the ADC.
After 12 6 2 (mean 6 standard deviation) days, all volunteers were imaged again (session 2) with the earlier described protocols at the same GE 1.5T MR system and the same software. As before, the ADC values for each DWI series were calculated automatically by the MR system and displayed as corresponding ADC maps. Data Analysis Quantitative analysis was performed at freestanding MR workstations by two radiology residents (J.Y.G., Y.L., with 4 and 2 years of experience in abdominal imaging, respectively), who were blinded to the subject identification information and imaging session. Region-of-interest (ROI) measurements were obtained in the pancreatic head, body and tail. Three nonoverlapping circular ROIs with a standardized size of 10 pixels were placed in homogeneous artifact-free areas, with large blood vessels excluded, on the images acquired without diffusion weighted gradient (b ¼ 0). By applying the copy and paste function of the workstation, ROIs were transferred to the same locations on the corresponding ADC maps (Fig. 2). The resulting mean ADC values for each of the three ROIs in the three corresponding anatomic locations represented the ADC values of pancreatic head, body and tail in a single individual. The ROIs were placed as consistent as possible for different scans. Statistical Analysis Statistical analysis was performed using SPSS 17.0 (SPSS Inc., Chicago, IL) and MedCalc Software (MedCalc, Mariakerke, Belgium). First, we assessed the agreement between observer 1 and 2 of ADC values measured at all three MR scanners. Second, we compared the ADC values between MR scanners measured by the same observer during imaging session 1.
ADC Reproducibility of the Pancreas Using DWI
1377
Figure 1. Representative images acquired from a volunteer. Diffusion-weighted images (b ¼ 600s/mm2) (a–c) and corresponding ADC maps (d–f) obtained at GE 1.5T, Siemens 1.5T, and Philips 3.0T.
Third, the ADC data measured at different sessions (only at GE 1.5T) by the same observer were also compared. Intraclass correlation coefficient (ICC) was used to assess the interobserver agreement. Pairedsample t-test was used for comparison of ADC values between scanners and imaging sessions. A P value of less than 0.05 was considered to be statistically significant. The significance level for comparison between scanners was 0.017(0.05/3) after Bonferroni correction. Furthermore, reproducibility of ADC measurements was determined as mean absolute difference (bias) and 95% confidence interval of the mean difference (limits of agreement) according to the Bland-Altman method (25). RESULTS The ADC values of pancreas showed good correlation between observer 1 and 2 at all three MR scanners (Table 2). During imaging session 1, the ADC values of pancreatic head, body and tail measured by the same observer at GE 1.5T were significantly higher than those at Philips 3.0T (P ¼ 0.008, 0.000, 0.000 for observer 1 and 0.008, 0.001, 0.000 for observer 2).
The ADC values of pancreatic head, body and tail measured at Siemens 1.5T were also significantly higher than those at Philips 3.0T (P ¼ 0.006, 0.000, 0.000 for observer 1 and 0.006, 0.002, 0.000 for observer 2). There were no significant differences in the measured ADC values between GE 1.5T and Siemens 1.5T (P ¼ 0.966, 0.394, 0.448 for observer 1 and 0.597, 0.691, 0.115 for observer 2) (Table 3). Figure 3 showed the results of Bland-Altman reproducibility analysis of ADC measurements between two scanners with 1.5T. The range of mean difference 6 limits of agreement (in 103mm2/s) between two scanners with 1.5T were 0.04–0.04 6 0.39–0.51 (observer 1), and 0.07– 0.02 6 0.43–0.53 (observer 2). There were no significant differences in the measured ADC values by the same observer between imaging session 1 and 2 at GE 1.5T (P ¼ 0.938, 0.464, 0.431 for observer 1 and 0.727, 0.151, 0.072 for observer 2) (Table 4). Figure 4 showed the results of Bland-Altman reproducibility analysis of ADC measurements between two imaging sessions. The range of mean difference 6 limits of agreement (in 103mm2/s) between two imaging sessions at GE1.5T were 0.03– 0.04 6 0.35–0.47 (observer 1), and 0.04–0.04 6 0.24–0.27 (observer 2).
Figure 2. Three ROIs (1–3) were placed in the pancreatic tail on the echo-planar image acquired without diffusion weighted gradient (b ¼ 0) (a), which were copied and pasted to the same locations on the corresponding ADC map (b).
1378
Ye et al.
Table 2 ADC Values of Pancreas Measured by Different Observers at Three MR Scanners* Scanner Location
GE1.5 head
body
SI1.5T tail
head
body
PH3.0 tail
head
body
tail
Observer1 2.24 6 0.33 2.01 6 0.23 1.88 6 0.26 2.24 6 0.23 2.04 6 0.26 1.84 6 0.28 2.06 6 0.23 1.80 6 0.23 1.56 6 0.25 Observer2 2.23 6 0.34 2.00 6 0.17 1.92 6 0.22 2.20 6 0.25 1.98 6 0.21 1.84 6 0.25 2.02 6 0.19 1.79 6 0.25 1.60 6 0.18 ICC 0.875 0.747 0.781 0.726 0.495 0.910 0.796 0.763 0.756 P value 0.000 0.000 0.000 0.000 0.006 0.000 0.000 0.000 0.000 *Data are mean values 6 standard deviation ( 103mm2/s). SI1.5T ¼ Siemens 1.5T; PH3.0T ¼ Philips 3.0T; head ¼ pancreatic head; body ¼ pancreatic body; tail ¼ pancreatic tail; ICC ¼ intraclass correlation coefficient.
DISCUSSION In clinical practice, the ADC values are often measured by different observers during follow-up examinations; hence, it might be helpful to know the interobserver difference of ADC measurements. Our results showed good agreement between observers in ADC measurements of pancreas, which indicate that ADC measurements have good reproducibility and can be performed by different observers. However, it is important to standardize the method when ADC is measured by different observers. In our study, there were significant differences in the measured ADC values of pancreas between GE 1.5T and Philips 3.0T, also between Siemens 1.5T and Philips 3.0T. The ADC values measured at both 1.5T scanners were significantly higher than those at 3.0T. In the previous studies (26,27), the ADC values of pancreas measured at 1.5T scanner were higher than at 3.0T but not significantly, perhaps related to the smaller n of these prior studies. However, the difference of pancreatic ADC between 1.5T and 3.0T in our study cannot be entirely attributed to difference in field strength, differences in echo time (TE) and vendor may be confounding factors. In our study, most parameters were kept the same across different MR scanners except TE. Because DWI is known to be error prone when signal-to-noise ratio (SNR) is poor, we adopted the shortest obtainable TE to ensure sufficient SNR for DW images. However, the shortest obtainable TE in SS-SE-EPI DWI sequence was different across scanners and it was the longest on Siemens 1.5T scanner, which may be due to the differences in pulse sequence design and gradient performance. Additionally, the TE setting in SS-SEEPI DWI sequence on GE 1.5T scanner cannot be
modified. Therefore, the chosen TE was not the same on different scanners in our study. In the previous studies (26,27), the MR systems of 1.5T and 3.0T came from same vendor. But, in our study, they are from different vendors. In biological tissue, diffusion may be restricted. The measured ADC can be affected by the diffusion time determined by DWI pulse sequence and imaging protocols as an indication of restricted diffusion (30). If the diffusion time is different in two DWI protocols, the measured ADC could be different due to restricted diffusion. In our study, it is difficult to determine the exact diffusion time for the twice-refocused SS-SE-EPI sequence on Siemens 1.5T scanner (31). Diffusion time was also different on the other two scanners (27.9 ms on GE 1.5T scanner, 20.8 ms on Philips 3.0T scanner). More research is needed to confirm the field strength effect without the influence of diffusion time. In our study, there were no significant differences in the measured ADC values of pancreas between GE 1.5T and Siemens 1.5T, and the Bland-Altman method indicated good reproducibility. The reproducibility of ADC measurements was good when measured at different scanners with same field strength. Because, in clinical practice, the follow-up examination for the same individual may be performed on a MR scanner with same field strength but from different vendors, it is necessary to know the ADC reproducibility measured at different scanners with same field strength. Our results indicate that ADC measurements have good reproducibility and the comparison of ADC values can be performed at different MR systems with the same field strength. In our study, there were no significant differences in the measured ADC values of pancreas between
Table 3 ADC Values of Pancreas Measured by the Same Observer at Three MR Scanners* GE1.5 PH3.0T Anatomic location 1
2
Pancreatic head Pancreatic body Pancreatic tail Pancreatic head Pancreatic body Pancreatic tail
GE1.5T 2.24 2.01 1.88 2.23 2.00 1.92
6 6 6 6 6 6
0.33 0.23 0.26 0.34 0.17 0.22
SI1.5T 2.24 2.04 1.84 2.20 1.98 1.84
6 6 6 6 6 6
0.23 0.26 0.28 0.25 0.21 0.25
PH3.0T 2.06 1.80 1.56 2.02 1.79 1.60
6 6 6 6 6 6
0.23 0.23 0.25 0.19 0.25 0.18
SI1.5T PH3.0T
GE1.5T SI1.5T
t value
P value
t value
P value
t value
P value
2.924 4.866 6.323 2.899 3.988 7.218
0.008 0.000 0.000 0.008 0.001 0.000
3.053 5.247 4.456 3.027 3.510 4.357
0.006 0.000 0.000 0.006 0.002 0.000
0.043 0.868 0.772 0.537 0.403 1.636
0.966 0.394 0.448 0.597 0.691 0.115
*Data are mean values 6 standard deviation ( 103mm2/s). SI1.5T ¼ Siemens 1.5T; PH3.0T ¼ Philips 3.0T; 1 ¼ observer 1; 2 ¼ observer 2.
ADC Reproducibility of the Pancreas Using DWI
1379
Figure 3. Reproducibility of pancreatic ADC measured at GE 1.5T and Siemens 1.5T scanners. Bland-Altman plots of difference of ADC measurements (y-axis) against mean ADC measurements (x-axis), with mean absolute difference (bias) (continuous line) and 95% confidence interval of the mean difference (limits of agreement) (dashed lines) for pancreatic head (a), pancreatic body (b), and pancreatic tail (c). Table 4 ADC Values of Pancreas Measured by the Same Observer in Different Time at GE1.5T* Anatomic location Pancreatic head Pancreatic body Pancreatic tail Pancreatic head 2 Pancreatic body Pancreatic tail 1
Session 1 2.24 2.01 1.88 2.23 2.00 1.92
6 6 6 6 6 6
0.33 0.23 0.26 0.34 0.17 0.22
Session 2 2.24 2.04 1.85 2.24 2.04 1.87
6 6 6 6 6 6
t value
0.32 0.078 0.25 0.744 0.22 0.801 0.34 0.354 0.15 1.484 0.21 1.889
P value 0.938 0.464 0.431 0.727 0.151 0.072
*Data are mean values 6 standard deviation ( 103mm2/s). 1 ¼ observer 1; 2 ¼ observer 2.
imaging session 1 and 2 at GE 1.5T, and the BlandAltman method indicated good reproducibility. The ADC measurements have good reproducibility measured at different sessions using the same MR scanner. The ADC values for each individuals were fairly stable over a period (up to 2 weeks), which was similar as shown in the previous studies (27,28). Rosenkrantz et al found that there was no significant difference in reproducibility of ADC for pancreas between 1.5T and 3.0T measured with a mean time interval of 12 days (27). Braithwaite et al found that there was no significant effect on ADCs of pancreas measured over a short-term (separated by five sequences within one imaging session) or
midterm (period between two imaging sessions with a mean time interval of 147 days) for a given individual (28). The ADC values of pancreatic head, body and tail measured in our study are different from those in the study by Dale et al (26). Dale et al measured the pancreatic ADC using a free-breathing technique at both 1.5T and 3.0T scanners, and indicated that the choice of b-values has a highly significant effect on ADC measurements with the general trend of decreasing measured ADC value with increasing b-value. With the b-value of 0 and 400, the mean ADC values of pancreatic head, body and tail were 2.65 103mm2/s, 2.89 103mm2/s, and 2.33 103mm2/s at 1.5T, 2.43 103mm2/s, 2.66 103mm2/s, and 2.24 103mm2/s at 3.0T. However, with the b-value of 0 and 800, the mean ADC values of pancreatic head, body and tail were 1.90 103mm2/s, 2.12 103mm2/s, and 1.75 103mm2/s at 1.5T, 1.71 103mm2/s, 1.79 103mm2/s, and 1.70 103mm2/s at 3.0T. The difference of pancreatic ADC between our study and Dale’s study may be due to the difference in b-values, and all results were mostly consistent with the effect of b-values on measured ADC value. Furthermore, the free-breathing technique adopted by Dale et al was substantially different from the respiratory-triggered technique in our study, which may also introduce some significant effects.
Figure 4. Reproducibility of pancreatic ADC measured at GE 1.5T with a time interval of 12 days. Bland-Altman plots of difference of ADC measurements (y-axis) against mean ADC measurements (x-axis), with mean absolute difference (bias) (continuous line) and 95% confidence interval of the mean difference (limits of agreement) (dashed lines) for pancreatic head (a), pancreatic body (b), and pancreatic tail (c).
1380
It is well known that more reliable calculations of ADC are possible if DWI sequences with more bvalues are used. However, we used a DWI sequence with only two b-values (0 and 600 s/mm2). This is because it is more typical in the current clinical practice. Additionally, only two b-values are available on our GE 1.5T scanner, so we used two b-values (0 and 600) for all three MR scanners to keep imaging parameters consistent. However, such choice for only two b-values may reduce confidence in the obtained ADC values, and the potential influence from the perfusion components could not be better characterized. Although abdominal DWI has been found in several studies to be robust and repeatable, our study indicates it is important to be careful in interpreting the change of measured ADC values. The difference in ADC values may be merely cause by the changes of MR scanner and/or measurement technique rather than the disease progression or treatment effect. There were several limitations in our study. First, the MR scanners were from different vendors in the study for assessing the effect of field strength on ADC measurement, so it was difficult to know whether the discrepancy of ADC values was due to one of the factors or a combination of both. Second, to keep the imaging parameters as consistent as possible, we performed all imaging examinations with the same scanning parameters except TE. This ideal scenario may not be achievable in actual clinical practice. Third, reproducibility beyond 2 weeks was not evaluated. The period of up to 2 weeks allows for the assessment of early response of treatment but not for long-term follow-up. Fourth, we did not evaluate the effect of different diffusion imaging techniques on the reproducibility of ADC measurement. It is known that the number of b values, the diffusion encoding direction, the breath modality, the use of parallel acquisition technique could affect the measured ADC value (4,25,26), so it is possible that they also affect reproducibility. Finally, our study population was relatively small and consisted of young, healthy subjects, agerelated effects could not be tested. Also what we really need to know is the reproducibility in clinical patients with disease, so further research is needed. In conclusion, the field strength of MR scanner may affect the measured ADC values of pancreas. The ADC values of pancreas show good reproducibility when measured at different MR systems with same field strength or over a period within 2 weeks at the same MR system. The specific results are only applicable to pancreatic ADC measurement using a similar DWI sequence with two b-values as in our study. REFERENCES 1. Lyng H, Haraldseth O, Rofstad EK. Measurement of cell density and necrotic fraction in human melanoma xenografts by diffusion weighted magnetic resonance imaging. Magn Reson Med 2000;43: 828–836. 2. Ichikawa T, Erturk SM, Motosugi U, et al. High-b-value diffusionweighted MRI in colorectal cancer. AJR Am J Roentgenol 2006; 187:181–184. 3. Koh DM, Takahara T, Imai Y, Collins DJ. Practical aspects of assessing tumors using clinical diffusion-weighted imaging in the body. Magn Reson Med Sci 2007;6:211–224.
Ye et al. 4. Koh DM, Collins DJ. Diffusion-weighted MRI in the body: applications and challenges in oncology. AJR Am J Roentgenol 2007; 188:1622–1635. 5. Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E, LavalJeantet M. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurological disorders. Radiology 1986;161:401–407. 6. Warach S, Chien D, Li W, Ronthal M, Edelman RR. Fast magnetic resonance diffusion-weighted imaging of acute human stroke. Neurology 1992;42:1717–1723. 7. Buckley BT, Wainwright A, Meagher T, Briley D. Audit of a policy of magnetic resonance imaging with diffusion-weighted imaging as first-line neuroimaging for in-patients withclinically suspected acute stroke. Clin Radiol 2003;58:234–237. 8. Chow LC, Bammer R, Moseley ME, Sommer FG. Single breathhold diffusion-weighted imaging of the abdomen. J Magn Reson Imaging 2003;18:377–382. 9. Ichikawa T, Haradome H, Hachiya J, Nitatori T, Araki T. Diffusion-weighted MR imaging single-shot echo-planar imaging in the upper abdomen: preliminary clinical experience in 61 patients. Abdom Imaging 1999;24:456–461. 10. Perman WH, Balci NC, Akduman I, Kuntz E. Magnetic resonance measurement of diffusion in the abdomen. Top Magn Reson Imaging 2009;20:99–104. 11. Shinya S, Sasaki T, Nakagawa Y, Guiquing Z, Yamamoto F, Yamashita Y. Usefulness of diffusion-weighted imaging (DWI) for the detection of pancreatic cancer: 4 cases reports. Hepatogastroenterology 2008;55:282–285. 12. Matsuki M, Inada Y, Nakai G, et al. Diffusion-weighed MR imaging of pancreatic carcinoma. Abdom Imaging 2007;32: 481–483. 13. Muraoka N, Uematsu H, Kimura H, et al. Apparent diffusion coefficient in pancreatic cancer: characterization and histopathological correlations. J Magn Reson Imaging 2008;27:1302–1308. 14. Momtahen AJ, Balci NC, Alkaade S, Akduman EI, Burton FR. Focal pancreatitis mimicking pancreatic mass: magnetic resonance imaging (MRI)/magnetic resonance cholangiopancreatography (MRCP) findings including diffusion-weighted MRI. Acta Radiol 2008;49:490–497. 15. Takeuchi M, Matsuzaki K, Kubo H, Nishitani H. High-b-value diffusion-weighted magnetic resonance imaging of pancreatic cancer and mass-forming chronic pancreatitis: preliminary results. Acta Radiol 2008;49:383–386. 16. Fattahi R, Balci NC, Perman WH, et al. Pancreatic diffusionweighted imaging (DWI): comparison between mass-forming focal pancreatitis (FP), pancreatic cancer (PC), and normal pancreas. J Magn Reson Imaging 2009;29:350–356. 17. Lee SS, Byun JH, Park BJ, et al. Quantitative analysis of diffusion-weighted magnetic resonance imaging of the pancreas: usefulness in characterizing solid pancreatic masses. J Magn Reson Imaging 2008;28:928–936. 18. Kartalis N, Lindholm TL, Aspelin P, Permert J, Albiin N. Diffusion-weighted magnetic resonance imaging of pancreas tumours. Eur Radiol 2009;19:1981–1990. 19. Kamisawa T, Takuma K, Anjiki H, et al. Differentiation of autoimmune pancreatitis from pancreatic cancer by diffusion-weighted MRI. Am J Gastroenterol 2010;105:1870–1875. 20. Inan N, Arslan A, Akansel G, Anik Y, Demirci A. Diffusionweighted imaging in the differential diagnosis of cystic lesions of the pancreas. AJR Am J Roentgenol 2008;191:1115–1121. 21. Fatima Z, Ichikawa T, Motosugi U, et al. Magnetic resonance diffusion-weighted imaging in the characterization of pancreatic mucinous cystic lesions. Clin Radiol 2011;66:108–111. 22. Erturk SM, Ichikawa T, Motosugi U, Sou H, Araki T. Diffusion weighted MR imaging in the evaluation of pancreatic exocrine function before and after secretin stimulation. Am J Gastroenterol 2006;101:133–136. 23. Akisik MF, Aisen AM, Sandrasegaran K, et al. Assessment of chronic pancreatitis: utility of diffusion-weighted MR imaging with secretin enhancement. Radiology 2009;250:103–109. 24. Kim H, Morgan DE, Buchsbaum DJ, et al. Early therapy evaluation of combined anti-death receptor 5 antibody and gemcitabine in orthotopic pancreatic tumor xenografts by diffusion-weighted magnetic resonance imaging. Cancer Res 2008;68:8369–8376. 25. Kwee TC, Takahara T, Koh DM, Nievelstein RA, Luijten PR. Comparison and reproducibility of ADC measurements in breathhold,
ADC Reproducibility of the Pancreas Using DWI respiratory triggered, and free-breathing diffusion-weighted MR imaging of the liver. J Magn Reson Imaging 2008;28:1141–1148. 26. Dale BM, Braithwaite AC, Boll DT, Merkle EM. Field strength and diffusion encoding technique affect the apparent diffusion coefficient measurements in diffusion-weighted imaging of the abdomen. Invest Radiol 2010;45:104–108. 27. Rosenkrantz AB, Oei M, Babb JS, Niver BE, Taouli B. Diffusionweighted imaging of the abdomen at 3.0 Tesla: image quality and apparent diffusion coefficient reproducibility compared with 1.5 Tesla. J Magn Reson Imaging 2011;33:128–135. 28. Braithwaite AC, Dale BM, Boll DT, Merkle EM. Short- and midterm reproducibility of apparent diffusion coefficient measure-
1381 ments at 3.0-T diffusion-weighted imaging of the abdomen. Radiology 2009;250:459–465. 29. Yoshikawa T, Kawamitsu H, Mitchell DG, et al. ADC measurement of abdominal organs and lesions using parallel imaging technique. AJR Am J Roentgenol 2006;187:1521– 1530. 30. Norris DG. The effects of microscopic tissue parameters on the diffusion weighted magnetic resonance imaging experiment. NMR Biomed 2001;14:77–93. 31. Reese TG, Heid O, Weisskoff RM, Wedeen VJ. Reduction of eddycurrent-induced distortion in diffusion MRI using a twicerefocused spin echo. Magn Reson Med 2003;49:177–182.
