(3.1.6)--10.头颈部肿瘤脑科学与影像新技术.pdf
Simple and Reliable Determination of IntravoxelIncoherent Motion Parameters for the DifferentialDiagnosis of Head and Neck TumorsMiho Sasaki,Misa Sumi,Sato Eida,Ikuo Katayama,Yuka Hotokezaka,Takashi Nakamura*Department of Radiology and Cancer Biology,Nagasaki University School of Dentistry,Nagasaki,JapanAbstractIntravoxel incoherent motion(IVIM)imaging can characterize diffusion and perfusion of normal and diseased tissues,andIVIM parameters are authentically determined by using cumbersome least-squares method.We evaluated a simpletechnique for the determination of IVIM parameters using geometric analysis of the multiexponential signal decay curve asan alternative to the least-squares method for the diagnosis of head and neck tumors.Pure diffusion coefficients(D),microvascular volume fraction(f),perfusion-related incoherent microcirculation(D*),and perfusion parameter that is heavilyweighted towards extravascular space(P)were determined geometrically(Geo D,Geo f,and Geo P)or by least-squaresmethod(Fit D,Fit f,and Fit D*)in normal structures and 105 head and neck tumors.The IVIM parameters were compared fortheir levels and diagnostic abilities between the 2 techniques.The IVIM parameters were not able to determine in 14 tumorswith the least-squares method alone and in 4 tumors with the geometric and least-squares methods.The geometric IVIMvalues were significantly different(p,0.001)from Fit values(+264%and 27624%for D and f values,respectively).Geo Dand Fit D differentiated between lymphomas and SCCs with similar efficacy(78%and 80%accuracy,respectively).Stepwiseapproaches using combinations of Geo D and Geo P,Geo D and Geo f,or Fit D and Fit D*differentiated betweenpleomorphic adenomas,Warthin tumors,and malignant salivary gland tumors with the same efficacy(91%accuracy=21/23).However,a stepwise differentiation using Fit D and Fit f was less effective(83%accuracy=19/23).Consideringcumbersome procedures with the least squares method compared with the geometric method,we concluded that thegeometric determination of IVIM parameters can be an alternative to least-squares method in the diagnosis of head andneck tumors.Citation:Sasaki M,Sumi M,Eida S,Katayama I,Hotokezaka Y,et al.(2014)Simple and Reliable Determination of Intravoxel Incoherent Motion Parameters for theDifferential Diagnosis of Head and Neck Tumors.PLoS ONE 9(11):e112866.doi:10.1371/journal.pone.0112866Editor:Sune N.Jespersen,Aarhus University,DenmarkReceived March 7,2014;Accepted October 20,2014;Published November 17,2014Copyright:?2014 Sasaki et al.This is an open-access article distributed under the terms of the Creative Commons Attribution License,which permitsunrestricted use,distribution,and reproduction in any medium,provided the original author and source are credited.Data Availability:The authors confirm that all data underlying the findings are fully available without restriction.All data underlying the findings in the presentstudy are freely available in the manuscript.Funding:The authors received the funds of this study from the internal source of Nagasaki University.The funders had no role in study design,data collectionand analysis,decision to publish,or preparation of the manuscript.Competing Interests:The authors have declared that no competing interests exist.*Email:takunagasaki-u.ac.jpIntroductionDiffusion occurs because of the non-ending movement of everysingle molecule 1.Brown first observed this phenomenon(although Ingenhousz found the phenomenon earlier than Browndid),and Einstein later gave this phenomenon a sound mathe-matical description considering a free diffusion process,where themolecules only collide with other molecules in a homogeneouscontainer without boundaries.Diffusion-weighted imaging(DWI)is based on MR signal attenuations caused by the displacement ofintracellular and extracellular water molecules for a given time.Inbiological tissues,however,the environment of water moleculescan hardly be called homogeneous:membranes,macromolecules,and fibers hamper the diffusion process 2.Furthermore,there isother incoherent motion within a voxel that can lead to signalattenuation;in particular,the water molecules in blood capillariesexhibit a pseudorandom motion in the tortuous capillaries.Le Bihan proposed that intravoxel incoherent motion(IVIM)imaging can distinguish between the pure molecular diffusion andmotion of water molecules in the capillary network with a singleDWI acquisition technique,provided that high b-values($200 s/mm2)and low b-values(,200 s/mm2)are used 3.The IVIMimaging can be characterized by 3 parameters:pure diffusioncoefficient(D);microvascular volume fraction(f);and perfusion-related incoherent microcirculation(D*)4.To determine theIVIM parameters from a multiexponential signal decay curve,theleast-squares method is usually used 4,5.However,the method iscumbersome,and thus may not be suitable for routine clinical use.Recently,some researchers have applied simplified methods fordetermining IVIM parameters to characterize tumors in the liver,prostate,and head and neck region 69.For example,Lewin etal shoed that the perfusion fraction parameter f determined byusing a geometric analysis of DW MR images can be a marker ofsorafenib treatment of patients with advanced hepatocellularcarcinoma 7.However,they did not indicate how precise thegeometric determination of the perfusion parameter comparedwith the conventional technique.In addition,Mazaheri et al notedthat the linear fit of the logarithmic signal using limited numbers ofb-value is statistically less appropriate than fitting the signals toexponential functions using a least-squares method 8.ThePLOS ONE|www.plosone.org1November 2014|Volume 9|Issue 11|e112866authors also suggested the importance of b-value selection used forthe simplified IVIM analysis.In simplified methods,the IVIMparameters are estimated by using a limited number 34 of b-values compared with the authentic IVIM imaging,which uses 913 b-values 4,5,10,11.However,the reliability in measurementsand effectiveness in diagnosing tumors with simplified IVIMtechniques using limited numbers of b-value has not been fullyinvestigated.Sasaki et al.reported the reproducibility of IVIMparameter measurements in evaluating the technique for func-tional assessment of the masticator muscles 12.However,therewas no published report that presented the reproducibility ofIVIM parameters in diagnosing tumors.In the present study,wedirectlycomparedtheIVIMparametervaluesthatweredeterminedbyasimplifiedgeometricmethodwiththosedetermined by the conventional least-squares method.We havealso compared the diagnostic accuracy for diagnosing head andneck squamous cell carcinomas(SCCs)and lymphomas as well asbenign and malignant salivary gland tumors between the 2methods.Materials and MethodsEthics statementThe Ethics Committee of Nagasaki University approved thisstudy.Informed consent was waived due to the retrospectiveTable 1.105 head and neck tumors.TumornBenign35Salivary gland tumorPleomorphic adenoma15Warthin tumor8Odontogenic tumorAmeloblastoma2Keratcysticodontogenic tumor2Odontogenic fibroma2Odontogenicmyxoma1Hemangioma1Angiomyoma1Myofibroma1Papiloma1Adenomatous goiter1Malignant70SCC25SCC node12Lymphoma14Salivary gland tumorCarcinoma ex.pleomorphic adenoma2Adenoid cystic carcinoma1Acinic cell carcinoma1Adenocarcinoma1Dedifferentiated carcinoma1Salivary duct carcinoma1Lymph node metastasis from malignant salivary gland tumor4Malignant melanoma1Nasopharyngeal carcinoma1Neuroendocrine carcinoma1Papillary thyroid carcinoma1Lymph node metastasis from papillary thyroid carcinoma3Ameloblastic carcinoma1Total105SCC,squamous cell carcinoma.Of 105 tumors,18 were excluded from the study owing to measurement errors,including 6 pleomorphic adenomas,2 lymphomas,2 SCCs(oropharynx andhypopharynx),1 SCC node,1 adenocarcinoma,1 metastatic node from papillary thyroid carcinoma,1 neuroendocrine carcinoma,1 ameloblastic carcinoma,1hemangioma,1 keratocystic odontogenic tumor,1 myxoma.doi:10.1371/journal.pone.0112866.t001Geometric Determination of IVIM ParametersPLOS ONE|www.plosone.org2November 2014|Volume 9|Issue 11|e112866牙源性肿瘤多形性腺瘤Figure 1.IVIM parameter determination by least-squares or geometric method.a,Least-squares method.Upper panel shows arepresentative signal decay curve obtained by using 11 b-values(0,10,20,30,50,80,100,200,300,400,800 s/mm2).At first step,D(Fit D)can beobtained by least-squares method using ln S200,ln S300,ln S400,and ln S800,and initial f value is calculated as S0Sinter=S0,where Sinteris theinterception of the logarithmic regression line obtained by using b-values of 200,300,400,and 800 s/mm2with the y-axis.Right panel showsrelationship between Sb/S0and varying b-values.Given D and initial f and D*values,f(Fit f)and D*(Fit D*)values can be obtained by least-squaresmethod based on the equation:Sb=S0 1f:exp bDzf:b DzD?.b,Geometric method.Graph shows geometric determination of IVIMparameters using 3(0,200,and 800 s/mm2)of the 11 b-values.D is calculated by the equation GeoD lnS200lnS800=600.f is estimated by theequation Geof S0Sinter=S0,and P is estimated by the equation GeoP lnS0lnSinter=200Geo P=(ln S0In Sinter)/200.c,Geometric methodbased on 4-b-value data.Graph shows geometric determination of IVIM parameters using 4(0,100,400,and 800 s/mm2)of the 11 b-values.D iscalculated by the equation GeoD4b lnS400lnS800=400,f is estimated by the equation Geof4b S0Sinter=S0,and P is estimated by theequation GeoP4blnS0lnSinter lnS100 lnSinter100:GeoD4b?fg=100.doi:10.1371/journal.pone.0112866.g001Table 2.Inter-and intraobserver errors in measuring Geo and Fit IVIM parameters.IVIM parameters%CVGeoFitInterobserver errorsD0.560.20.560.3f5.260.89.767.8P/D*5.160.816.2611.0Intraobserver errorsD0.960.91.060.9f5.465.411.467.9P/D*5.565.5a19.768.6a%CV,percent coefficient of variation;Geo,geometric measurement;Fit,least-squares method.P/D*,Geo P/Fit D*.a,significant difference between Geo and Fit values(p=0.0195,Wilcoxon signed-rank test).doi:10.1371/journal.pone.0112866.t002Geometric Determination of IVIM ParametersPLOS ONE|www.plosone.org3November 2014|Volume 9|Issue 11|e112866nature of the study.Patient records/information was anonymizedand de-identified prior to analysis.PatientsWe retrospectively studied DW MR images of patients withheadandnecktumorswhounderwentpreoperativeMRexaminations between March 2003 to April 2012.We selectedhead and neck tumors from patients(1)who underwent diffusion-weighted MR imaging as well as conventional contrast-enhancedand non-enhanced T1-weighted and fat-suppressed T2-weightedMR imaging;(2)whose tumors were excised and histologicallyproven;and(3)whose DW images were good in quality withoutany severe susceptibility artifacts that would interfere with IVIManalysis.Consequently,the study cohort included 105 head andneck tumors(35 benign and 70 malignant tumors)that arose in 94patients(56 men and 38 women;average age,62615 years;agerange,391 years).Detailed tumor pathology is listed in Table 1.DW MR images of the healthy parotid glands(n=21)and themasseter muscles(n=21)of the contralateral sides in patients withparotid tumor were also analyzed for comparing the IVIMparameters determined by using geometric or least-squaresmethods.MR imagingMR imaging was performed using a 1.5-T MR unit(GyroscanIntera 1.5T Master;Philips Healthcare,Best,The Netherlands).73 patients were scanned by using a 2-channel 17-cm614-cm(Synergy-Flex M),7 patients by using a 2-channel 20-cm(Synergy-Flex L)surface coil,and 14 patients by using a 3-channel head andneck coil(Synergy Head Neck).T1-and T2-weighted MR imagingWe obtained axial T1-and fat-suppressed(spectral attenuatedwith inversion recovery,SPAIR)T2-weighted MR images(TR/TE/numberofsignalacquisitions=500 ms/15 ms/2and6385 ms/80 ms/2,respectively)by using a turbo spin-echo(TSE)sequence(TSE factor=3 and 15,respectively).We used a200-mm FOV,2566204 scan and 5126512 reconstruction matrixsizes,a 4-mm slice thickness and a 0.4-mm slice gap.For contrast-enhanced T1-weighted MR imaging,a gadolinium-based agent(gadopentatatedimeglutimine,Magnevist;BayerHealthcare,Berlin,Germany)was intravenously injected at a dose of 0.2 mLper kg of body weight and a rate of 1.5 mL/s.DW MR imagingAxial DW images(TR/TE=1625 ms/81 ms)were obtainedusing single-shot,spin-echo echo planar imaging(SE-EPI).TheEPI factor was 47,and Sensitivity Encoding(SENSE)factor was 2.We used a 200-mm FOV,4-mm slice thickness,0.4-mm slice gap,and 112690 matrix size.The measured pixel size was 1.79/2.28/4 mm.We used 11 b-values(0,10,20,30,50,80,100,200,300,400,and 800 s/mm2).The total acquisition time was 1 min 53 sper 5 slices.Regions of interestA region of interest(ROI)was manually placed onto each tumorarea such that it encompassed as much of the tumor area aspossible.The mean ROI area was 3.462.8 cm2(0.818.1 cm2).Visually large cystic or necrotic areas were excluded from thepresent analysis.We used the contrast-enhanced T1-weighted andfat-suppressed T2-weighted MR images as references to determinetumor areas on the corresponding DW images.We compared theIVIM values between geometric and least-squares methods basedon the IVIM values calculated from ROI-averaged signalintensities.We used DW image slices including the 13 maximaltumor areas,and the IVIM values obtained from each ROIs wereaveraged.For the healthy parotid glands and masseter muscles,irregular ROIs were placed so that they included as much of thegland or muscle area as possible,but did not include large vessels,such as the retromandibular vein,or intraglandular main ducts.Aradiologist with 20-year experience in head and neck radiologyplaced ROIs and analyzed IVIM images.Figure 2.IVIM parameters of normal structures and tumors in the head and neck region.Plot graphs show D(Geo D),f(Geo f),and P(GeoP)values that were determined by geometric method;and D(Fit D),f(Fit f),and D*(Fit D*)values that were determined by least-squares method ofnormal structures(parotid glands,open circles;and masseter muscles,open squares)and head and neck tumors(closed circles).Broken whitecontours indicate tumor areas.Parotid gland:Geo D,Geo f and Geo P=0.7660.1761023mm2/s,0.2060.04,and 1.1260.2761023mm2/s,respectively;and Fit D,Fit f,and Fit D*=0.7560.1661023mm2/s,0.2060.05,and 62.96646.7861023mm2/s,respectively.Masseter muscle:Geo D,Geo f,and Geo P=0.9960.5161023mm2/s,0.2460.10,and 1.4160.7161023mm2/s,respectively;and Fit D,Fit f,and Fit D*=0.9660.5161023mm2/s,0.2560.10,and 40.50630.1361023mm2/s,respectively.Tumors:Geo D,Geo f,and Geo P=1.0060.3861023mm2/s,0.1160.08,and0.6160.4861023mm2/s,respectively;Fit D,Fit f,and Fit D*=0.9960.3761023mm2/s,0.1260.08,and 24.14621.1561023mm2/s,respectively.Insert,Geo P distribution on a small scale.The values are the results of integrated signal intensities within the ROIs.*,p,0.001(Wilcoxon signed-ranktest).doi:10.1371/journal.pone.0112866.g002Geometric Determination of IVIM ParametersPLOS ONE|www.plosone.org4November 2014|Volume 9|Issue 1
