TECHNICAL NOTEA comparative study between arterial spin labeling and CT perfusion methods on hepatic portal venous flowYoshiaki Katada •Toshiro Shukuya •Miho Kawashima •Miwako Nozaki •Hiroshi Imai •Takeshi Natori •Masaya TamanoReceived:16April 2012/Accepted:20August 2012ÓJapan Radiological Society 2012AbstractPurpose The purpose of this study was to evaluate the feasibility and potential usefulness of unenhanced mag-netic resonance (MR)hepatic portal perfusion using arte-rial spin labeling (ASL)among healthy volunteers and hepatocellular carcinoma patients.Materials and methods The five healthy volunteers underwent unenhanced MR perfusion with inversion time 2(TI2)values at 500-ms intervals between 2,000and 4,000ms,and the 12patients underwent unenhanced MR perfusion using ASL and computed tomography (CT)perfusion during superior mesenteric artery (SMA)por-tography.The regions of interest were placed in both the right and left lobes of the liver or both the right anterior and posterior segments of the liver and were placed over the tumor if a lesion was located within a particular perfusion study slice.Results In the healthy volunteer study,perfusion rate in hepatic parenchyma showed a peak at the TI2value of 3,000ms (254.3ml/min/100g ±58.3).In patients,a fair correlation was observed between CT and MR perfusion (r =0.795,P \0.01).Conclusion Our results demonstrate a significant fair correlation between unenhanced MR hepatic portal perfu-sion imaging using ASL and CT perfusion during SMA portography.Keywords Arterial spin labeling ÁLiver portal perfusion ÁUnenhanced MRI ÁCT perfusion ÁMR perfusionIntroductionHepatic blood flow has been evaluated using various methods based on advances in imaging modalities,such as ultrasonography (US),computed tomography (CT),and magnetic resonance imaging (MRI)[1–3].Various meth-ods of independently investigating portal and hepatic arterial blood flow in the liver have been studied [1–3].CT perfusion imaging during superior mesenteric arterial (SMA)portography and hepatic arteriography can be used to quantify pure arterial and portal blood perfusion [4,5],although these examinations are highly invasive.Unen-hanced MR perfusion imaging using the arterial spin labeling (ASL)technique was introduced to quantify per-fusion in the brain [6–8],and several researchers have already reported that unenhanced MR perfusion using ASL is a promising tool for noninvasive estimation of perfusion in various organs,including brain and kidney [9–11].The portal vein shows relatively slow flow,and in normal breathing synchronization timing,labeled portal blood does not reach hepatic parenchyma.For accurate quantificationY.Katada (&)ÁT.Shukuya ÁM.Kawashima ÁM.Nozaki Department of Radiology,Dokkyo Medical University Koshigaya Hospital,2-1-50,Minami-Koshigaya,Koshigaya,Saitama 343-8555,Japan e-mail:yoshiaki@dokkyomed.ac.jpH.ImaiSiemens Japan K.K.,Takanawa Park Tower,3-20-14,Higashi-Gotanda,Shinagawa-ku,Tokyo 141-8644,Japan T.NatoriSecond Department of Surgery,Dokkyo Medical University Koshigaya Hospital,2-1-50,Minami-Koshigaya,Koshigaya,Saitama 343-8555,JapanM.TamanoDepartment of Gastroenterology,Dokkyo Medical University Koshigaya Hospital,2-1-50,Minami-Koshigaya,Koshigaya,Saitama 343-8555,JapanJpn J RadiolDOI 10.1007/s11604-012-0127-yof portal perfusion,we introduced and evaluated a two-respiration interval method.Only a few reports have been published of unenhanced MR perfusion imaging of the liver using ASL[12,13],and pure hepatic portal perfusion imaging has not been previously reported.Hence,the purpose of this study was to evaluate the feasibility and potential usefulness of unenhanced MR portal perfusion imaging using a1.5-T device and the ASL technique by comparing results with CT perfusion imaging during SMA portography and to test the data acquisition technique during a two-respiration interval.Materials and methodsHealthy volunteer study for reproducibility assessmentIn September2010,unenhanced MR portal perfusion imaging studies were performed infive healthy male vol-unteers(mean age30years).All were instructed to breathe normally during the scanning.Sequence parameters are described below in the‘‘MRI protocol and application of unenhanced MR portal perfusion imaging’’section.For the purpose of determining the optimal inversion time2(TI2) for measuring hepatic portal perfusion,a series of TI2 values at500-ms intervals from2,000to4,000ms were used,and repetition time(TR)and echo time(TE)were minimum values for all TI2values.Patient study for comparison with CT perfusionduring SMA portography in patients with liverfibrosisBetween October2010and December2010,unenhanced MR portal perfusion imaging studies were performed in30con-secutive patients.This study was conducted in accordance with the ethical standards of the World Medical Association (Declaration of Helsinki).Institutional review board approval and written informed consent from all the patients were obtained before patient registration.All patients were instructed to breathe normally during the scanning.Of these 30patients,12underwent CT perfusion during SMA por-tography before transcatheter arterial chemoembolization for hepatocellular carcinoma(HCC),and these12patients(nine men,three women)with a mean age of66.5(range 53–81)years were enrolled.All patients had HCC with liver cirrhosis(Child-Pugh A in nine and B in three),and the cause of chronic liver disease was viral in ten(hepatitis type C in eight and type B in two)and unknown in two patients. MRI protocol and application of unenhanced MR portal perfusion imagingEach examination was conducted with the individual in the supine position;a 1.5-T MRI system(MAGNETOM Avanto;Siemens Medical Systems,Erlangen,Germany) equipped with a phased-array coil was used for all exam-inations.MRI sequence for ASL was as follows:syngo ASL imaging application software,quantitative imaging of perfusion using a single subtraction,and second version (QUIPPS II)with thin-slice TI1periodic saturation (Q2TIPS)[6,14].This ASL imaging application software was adopted as a single-compartment model,and perfusion calculations were performed with an assumed T1of blood (T1B)of1,200ms.Signal difference D M(control-tag)is proportional to bloodflow in the below equation:D M¼M0BÁfÁTI1ÀM0Bð1À2expÀTI2/T1B½ ÁqÞfÁTI1 D M¼2M0BÁfÁTI1ÁexpÀTI2=T1B½ Áqand relative bloodflow was summarized by the following equation:f¼½D M=M0B exp TI2=T1BðÞ=2tÁqwhere M0B is fully relaxed magnetization of arterial blood, f is the bloodflow in milliliters of blood/milliliters of issue/ minute,q is a correction factor that accounts for a shift in the T1decay of the tag due to exchange of tagged spins from blood into tissue,and t is duration time of tag.The ASL parameters were as follows:2D echo-planar imaging (EPI)with a respiration trigger;TR)/TE=3,303.4(mini-mum)/21ms;field of view(FOV)380mm975.4% (286.7mm);104986imaging matrix; 3.793.7mm voxel size/10-mm slice thickness(four slices);distance factor20%;flip angle(FA)90°;fat suppression with a chemical-shift selective(CHESS);TI1=1,000ms;satu-ration stop time=2,000ms;TI2=3,000ms;flow limit 5cm/s;measurement,31times;parallel imaging(iPAT) generalized autocalibrating partially parallel acquisitions (GRAPPA);EPI factor78;bandwidth2,090Hz/pixel;echo spacing0.58ms;data acquisition time1min52s.The labeling area was defined as an oblique rectangular area covering the mesenteric and portal venous area,and the celiac trunk was excluded to avoid mislabeling as hepatic arterial blood.In patients with anomalous right hepatic artery originating from SMA,the labeling area was set on the distal side to the right hepatic artery to avoid misla-beling.This labeling area is shown diagrammatically in Fig.1a,and a respiration trigger image is shown in Fig.1b.CT perfusion during SMA portography and CTAP protocolsA4-F catheter(CHC-A;Terumo-Clinical Supply,Gifu, Japan)was inserted into the SMA.CT perfusion during the SMA portography protocol consisted of20ml of diluted contrast medium[10ml of iopamidol(Iopamiron300; Bayer,Germany)and10ml of sterilized saline solution]Jpn J Radiolinjected at a rate of 5ml/s via a catheter placed in the SMA.CT images were acquired during the 50-s period starting immediately after the beginning of contrast med-ium injection.During the procedure,all patients received nasal oxygen inhalation at 4L/min to prevent motion artifacts arising from insufficient breath-holding [5].Three-slice cine CT images were acquired at 120kV,100mA,and 1.0s/rev,followed by image reconstruction at 1.0-s intervals and a slice thickness of 8mm.Data were trans-ferred to a workstation for analysis using commercially available CT perfusion analysis software (syngo Body Perfusion;Siemens),which uses the maximum-slope method to analyze hepatic blood flow.All contrast medium injections were performed using a power injector.Statistical analysisTo evaluate reproducibility of this perfusion technique,intraclass correlation coefficients of the variable TI2values were calculated to estimate blood flow values in the five healthy volunteers.Regions of interest (ROIs)were placedin almost the same area for all TI2values.To determine the relationship between pure portal blood flow data for CT perfusion and the same for MR perfusion using ASL in patients,ROIs were placed in almost the same area between CT and MR perfusion studies,and this area encompassed the liver tumor (if the tumor was located in the perfusion study slices).ROIs were placed in liver parenchyma in both right and left lobes or in both right anterior lobe and posterior lobe to avoid blood vessels.As a rule,ROIs of he liver parenchyma were at least 100mm 2in size.The significance and strength of relationships examined in this study were expressed using the Pearson correlation coefficient (r )and regression line slope.Agreement between different perfusion methods was assessed with the Bland–Altman method [15].Statistical analyses were performed using a commercially available software package (IBM SPSS Statistics version 19;IBM Inc.).A p value \0.05was regarded as significant.ResultsMR portal perfusion imaging was successfully obtained in all healthy volunteers and patients,with no adverse events or technical failures during image acquisitions (Fig.2a,b),and CT perfusion during SMA portography was also suc-cessfully performed (Fig.2c).Among the healthy volun-teers,hepatic parenchyma perfusion rate was very high [3,000(mean 254.3ml/min/100g ±58.3)or 3,500(230.8ml/min/100g ±93.0)ms](Fig.3).At shorter TI2values,the portal branch itself exhibited very high signal intensity and the hepatic parenchyma relatively low signal intensity.This means that labeled portal blood reached the portal branch but not the hepatic parenchyma,and an accurate assessment of portal perfusion was not achieved using shorter TI2values.Longer TI2values tended to result in higher blood flow measurement,and the standard deviation (SD)also tended to increase at TI2values of C 3,500ms.Blood flow data for variable TI2was signifi-cant (P \0.05).The error bar graph is shown in Fig.3.In the patient study,correlation between the pure liver portal blood flow data obtained using CT perfusion and that using MR perfusion was significant.A simple logistic regression model yielded a significant linear regression (r =0.795,P \0.01),as shown in Fig.4a.A strong cor-relation (r =0.672,P \0.01)was also found in a sub-group of ROIs without an HCC lesion (n =24),as shown in Fig.5a.According to Bland–Altman plot analysis,mean differences and limits of agreement were 83.80ml/min/100g (limits of agreement -78.78,237.82)for all ROIs,as shown in Fig.4b,and 114.76ml/min/100g (limits of agreement 0.03,229.49)for ROIs without HCC,as shown in Fig.5b.This analysis shows an overestimationofFig.1a Labeling area (shaded area ).Celiac trunk was excluded from the labeling area.b Respiratory trigger display beling pulse occurred at the onset of the first exhale phase;data was acquired during the late phase of the second exhale phaseJpn J Radiolhepatic portal perfusion volumes using MR perfusion compared with hepatic portal perfusion volumes using CT,and the variability of the difference is decreased with increasing magnitude of measurements.DiscussionMR perfusion is usually performed using the dynamic susceptibility contrast (DSC)method,which involves contrast medium [1,4].However,following introduction of the ASL method [6–14],MR perfusion can be performed without the use of a contrast medium,but only parameters reflecting relative blood flow can usually be obtained.In comparison with CT perfusion,MR perfusion using ASL can be performed without radiation exposure,but the latter method is not yet capable of accurate quantifications that take into account the effectiveness of transit time.MR perfusion using the Look-Locker sequence,known as quantitative signal targeting by alternating radiofrequency pulses (STAR)labeling of arterial regions (QUASAR),enables quantification of cerebral blood flow using a deconvolution method [16].The liver has a dual blood supply comprising hepatic arterial and portal flow,and various methods have been proposed to visualize these two blood supplies separately.However,only CT perfusion during SMA portography and during hepatic arteriography are capable of accurately analyzing pure portal and arterial blood perfusion of the liver [4,5].These methods are highly invasive compared with CT or MR perfusion using contrast medium via an intravenous injection [1–3].Furthermore,unlike measuring brain perfusion,measuring liver perfusion can be influ-enced by respiration,making a good signal-to-noise ratio (SNR)difficult to obtain [17].In ASL,water in the blood itself is used as an endoge-nous tracer,allowing perfusion assessments without the risk of nephrotoxicity.One drawback of ASL,however,is the relatively long acquisition time due to patientbreathingFig.2A 78-year-old man with hepatocellular carcinoma compli-cated by mild liver cirrhosis.a Blood flow map of magnetic resonance (MR)perfusion using arterial spin labeling (ASL).b True fast imaging with steady-state precession (FISP)image of the same slice of the MR perfusion map.c Blood flow map of computed tomography (CT)perfusion during superior mesenteric artery (SMA)portography.CT perfusion image shows an axial plane;MR perfusion and True-FISP images show an oblique axialplaneFig.3Data from five healthy volunteers.An inversion time 2(TI2)value of 3,000ms resulted in the highest hepatic portal flow;a TI2value of 3,500ms resulted in a wider variation of valuesJpn J Radioland the relatively poor spatial resolution compared with contrast-enhanced perfusion studies.Nevertheless,the fact that perfusion studies can be performed without using contrast media is a major advantage of ASL.Portal blood flow is relatively slow and steady,with a value of about 20cm/s [18],which has been a major obstacle to per-forming accurate portal perfusion studies.Portal blood labeled within mesenteric and splenic veins reached the main portal branch about 1,000–1,200ms after the labeling tag pulse and was perfused to liver parenchyma within a few seconds [18–20].This slow and steady portal blood flow led to a long TI2time,and this long TI2made it difficult to obtain complete data acquisition during one respiration interval.In our study,we introduced data acquisition during an interval of two respirations to cope with the long TI2time.The introduction of this two res-piration interval technique enabled good liver perfusion of the labeled portal flow.The TI2value of 3,000ms used in this study required a TR value of 3,300ms,and this long TR value required data acquisition over tworespirationFig.4a Correlation between liver portal blood flow data obtained using computed tomography (CT)perfusion and magnetic resonance (MR)perfusion for all regions of interest (ROIs).Dotted lines show 95%confidential intervals (CI).b Bland–Altman plot of agreement between CT and MR perfusion for all ROIs.Dashed lines represent mean differences ±2standard deviations (limits of agreement -78.78,237.82)Fig.5a Correlation between liver portal blood flow data obtained using computed tomography (CT)perfusion and magnetic resonance (MR)perfusion for regions of interest (ROIs)without hepatocellular carcinoma (HCC).Dotted lines show 95%confidential intervals (CI).b Bland–Altman plot of agreement between CT and MR perfusion for ROIs without HCC.Dashed lines represent mean differences ±2standard deviations (limits of agreement 0.03,229.49)Jpn J Radiolintervals.Our methods reduced the measurement value to 31times,compared with an original of81times,to reduce respiratory motion artifacts.According to results,our method enabled unenhanced portal perfusion images of people with healthy livers and patients with mild hepaticfibrosis to be obtained,and a fair correlation(r=0.795,P\0.01)was observed between unenhanced MR perfusion and CT perfusion during SMA pared with previous methods of liver perfusion,our method has several advantages:First,it does not require contrast media.Second,although accurate quantification remains difficult,only this ASL method using a two-respiration interval technique can obtain a pure liver portal perfusion image,compared with CT and MR perfusion with contrast media via an intravenous injection. The low spatial resolution perfusion image obtained using our method is comparable with that obtained using CT perfusion;however,the perfusion defect areas of the liver tumor can be evaluated in patients with mild to moderate liverfibrosis(Child–Pugh class A and B).In patients with severe liverfibrosis,such as Child–Pugh C,however,our method of obtaining liver portal perfusion images had a very low success late(data not shown).Portal bloodflow and volume of the severefibrotic liver tissue was remark-ably reduced compared with values for healthy liver,and mean transit time was also greatly extended.These factors made accurate evaluations of liver portal perfusion diffi-cult,as a longer TI2time was required and the labeled blood signals were greatly weakened,resulting in a poor SNR.Our study had several limitations:First,the cohort was small,evaluating two groups:healthy volunteers and HCC patients with mild to moderate liverfibrosis.Second,CT perfusion used the maximum-slope method,and various methods used to quantify liver perfusion are controversial.A dual-input one-compartment model method was pro-posed by Materne et al.[21],and significant linear corre-lations were observed between perfusion parameters obtained in the maximum-slope and dual-input one-com-partment model methods except for hepatic arterial perfu-sion with arterial injection[22].Kanda et al.[2]showed the mean hepatic portal perfusion with venous bolus injection using the maximum-slope method was significantly lower than that of the dual-input one-compartment model,and we may have underestimated hepatic portal bloodflow. However,our CT perfusion during arterial portography was adopted using intra-arterial bolus injection via SMA, and the bolus of the portalflow was sharp compared with an aortic injection[22].Third,the study was conducted as a technical development research,and various parameters of our ASL application software were suitable only for brain perfusion study and adopted a single-compartment model.A two-compartment exchange model for perfusion quantification using ASL is now available,which corrects for the assumption that the capillary has infinite perme-ability to water[23,24].As the T1of blood is longer than the T1of tissue,signal decay will be slower than predicted by the single-compartment model,causing perfusion overestimation[23,24].We adjusted some parameters to make the method more suitable for liver portal perfusion studies,and according to the Bland–Altman plot analysis, our unenhanced perfusion method may contain systematic error.In the brain,perfusion is overestimated by approxi-mately60%in white matter for typical human perfusion rate at1.5T with a measurement time of3s using the single-compartment model.T1value of liver parenchyma was about600ms at1.5T[25],and this value gives sig-nificant overestimation because there is a larger difference between blood and liver parenchyma T1values[23,24],as is the case with white matter.However,T1values of liver are close to the T1value of white matter,and this may be one advantage for determining whether the application for the brain may be applicable to that of liver parenchyma. Although the development of our method is ongoing and in was necessary to adjustment various parameters,further study and development of MR imaging technologies will resolve this problem.Our perfusion method may require further optimization to ensure that quantitative data on pure hepatic portal bloodflow is accurate.In the brain,quanti-tative data on cerebral bloodflow can be obtained,but similar mathematical calculations and assumptions are not applicable to liver portal perfusion studies.However,the pure portal bloodflow data obtained using CT and MR perfusion were significantly correlated.Forth,a TI2value of3,000ms weakened blood labeling compared with results in brain perfusion studies.However,the magnitude of liver portal perfusion was much higher than that for brain perfusion,and signal intensity was also higher than that for brain perfusion.Finally,our method did not con-sider the effectiveness of transit time.As mentioned earlier, respiratory liver movement has been a major barrier to the introduction of the Look-Locker sequence,taking into account the focal transit time difference.In this study,liver portal perfusion was influenced by TI2value,but portal perfusion data for CT and MR perfusion using a TI2value of3,000ms exhibited a fair linear correlation.Despite the limitations,our study demonstrated that unenhanced MR portal perfusion imaging using ASL and a two-respiratory interval technique in individuals with healthy livers and HCC patients with mild to moderate hepaticfibrosis was significantly correlated with the results for CT perfusion during SMA portography.This perfusion method,which does not require the use of contrast media, is a noninvasive MR perfusion technique that may offer great potential as an alternative imaging method for pure liver portal perfusion.Jpn J RadiolAcknowledgments We thank Tsubasa Kaji,Siemens Japan K.K., for technical assistance.References1.Pandharipandle PV,Krinsky GA,Rusinek H,Lee VS.Perfusionimaging of the liver:current challenges and future goals.Radi-ology.2005;234:661–73.2.Kanda T,Yoshikawa T,Ohno Y,Kanata N,Koyama H,NogamiM,et al.Hepatic computed tomography perfusion:comparison of maximum slope and dual-input single-compartment methods.Jpn J Radiol.2010;28:714–9.3.Martirosian P,Boss A,Schraml C,Schwenzer NF,Graf H,Claussen CD,et al.Magnetic resonance perfusion imaging without contrast media.Eur J Nucl Med Mol Imaging.2010;37: S52–64.4.Komemushi A,Tanigawa N,Kojima H,Kariya S,Sawada S.CTperfusion of the liver during selective hepatic arteriography:pure arterial blood perfusion of liver tumor and parenchyma.Radiat Med.2003;21:246–51.5.Kojima H,Tanigawa N,Komemushi A,Kariya S,Sawada S.Computed tomography perfusion of the liver:assessment of pure portal bloodflow studied with CT perfusion during superior mesenteric arterial portography.Acta Radiol.2004;45:709–15.6.Luh WM,Wong EC,Bandettini PA,Hyde JS.QUIPPS II withthis-slice TI1periodic saturation:a method for improving accu-racy of quantitative perfusion imaging using pulsed arterial spin labeling.Magn Reson Med.1999;41:1246–54.7.van Laar PJ,van der Grond J,Hendrikse J.Brain perfusion ter-ritory imaging:methods and clinical applications of selective arterial spin-labeling MR imaging.Radiology.2008;246:354–64.8.Weber MA,Thilmann C,Lichy MP,Gu¨nther M,Delorme S,Zuna I,et al.Assessment of irradiated brain metastases by means of arterial spin-labeling and dynamic susceptibility-weighted contrast-enhanced perfusion MRI:initial results.Invest Radiol.2004;39:277–87.9.Schraml C,Schwenzer NF,Martirosian P,Claussen CD,SchickF.Perfusion imaging of the pancreas using an arterial spinlabeling technique.J Magn Reson Imaging.2008;28:1459–65.nzman RS,Wittsack HJ,Martirosian P,Zgoura P,Bilk P,Kro¨pil P,et al.Quantification of renal allograft perfusion using arterial spin labeling MRI:initial results.Eur Radiol.2010;20:1485–91.11.Bazelaire CD,Rofsky NM,Duhamel G,Michaelson MD,GeorgeD,Alsop D.Arterial spin labeling bloodflow magnetic resonance imaging for the characterization of metastatic renal cell carci-noma.Acad Radiol.2005;12:347–57.12.Gash HM,Li T,Lopez-Talavera JC,Kam AW.Liver perfusionMRI using arterial spin labeling.(abstr).In:Proceedings of the 10th Annual Meeting of ISMRM.Honolulu,HI:International Society of Magnetic Resonance in Medicine;2002.p.1939.13.Hoad C,Costigan C,Marciani L,Kaye P,Spiller R,Gowkand P,et al.Quantifying bloodflow and perfusion in liver tissue using phase contrast angiography and arterial spin labeling.(abstr)In: Proceedings of the19th Annual Meeting of ISMRM.Montreal, Canada:International Society of Magnetic Resonance in Medi-cine;2011.p.794.14.Wong EC,Buxton RB,Frank LR.Quantitative imaging of per-fusion using a single subtraction(QUIPSS and QUIPSS II).Magn Reson Med.1998;39:702–8.15.Bland JM,Altman DG.Statistical methods for assessing agree-ment between two methods of clinical ncet.1986;327:307–10.16.Petersen ET,Lim T,Golay X.Model-free arterial spin labelingquantification approach for perfusion MRI.Magn Reson Med.2006;55:219–32.17.Aruga T,Itami J,Aruga M,Nakajima K,Shibata K,Nojo T,et al.Target volume definition for upper abdominal irradiation using CT scans obtained during inhale and exhale phases.Int J Radiat Oncol Biol Phys.2,000;48:465–9.18.Sugimoto H,Kaneko T,Hirota M,Inoue S,Takeda S,Nakao A.Physical hemodynamic interaction between portal venous and hepatic arterial bloodflow in humans.Liver Int.2005;25:282–7.19.Shimada K,Isoda H,Okada T,Kamae T,Arizono S,Hirokawa Y,et al.Non-contrast-enhanced MR portography with time-spatial labeling inversion pulses:comparison of imaging with three-dimensional half-fourier fast spin-echo and true steady-state free-precession sequences.J Magn Reson Imaging.2009;29:1140–6.20.Shimada K,Isoda H,Okada T,Kamae T,Arizono S,Hirokawa Y,et al.Unenhanced MR portography with a half-fourier fast spin-echo sequence and time-space labeling inversion pulses:pre-liminary results.Am J Roentgenol.2009;193:106–12.21.Materne R,Van Beers BE,Smith AM,Leconte I,Jamart J,De-houx JP,et al.Non-invasive quantification of liver perfusion with dynamic computed tomography and a dual-input one-compart-ment model.Clin Sci(Lond).2,000;99:517–25.22.Miyazaki M,Tsushima Y,Miyazaki A,Paudyal B,Amanuma M,Endo K.Quantification of hepatic arterial and portal perfusion with dynamic computed tomography:comparison of maximum-slope and dual-input one-compartment model methods.Jpn J Radiol.2009;27:143–50.23.Parkes LM,Tofts PS.Improved accuracy of human cerebralblood perfusion measurements using arterial spin labeling: accounting for capillary water permeability.Magn Reson Med.2002;48:27–41.24.Parkes LM.Quantification of cerebral perfusion using arterialspin labeling:two compartment models.J Magn Reson Imag.2005;22:732–6.25.de Bazelaire CM,Duhamel GD,Rofsky NM,Alsop DC.MRimaging relaxation times of abdominal and pelvic tissues measured in vivo at 3.0T:preliminary results.Radiology.2004;230:652–9.Jpn J Radiol。