ContentslistsavailableatScienceDirect
NeuroImage
journalhomepage:www.elsevier.com/locate/neuroimage
Structural connectivity-based segmentation of the human entorhinal cortex
Ingrid Framås Syversen
a,∗, Menno P. Witter
a, Asgeir Kobro-Flatmoen
a, Pål Erik Goa
b, Tobias Navarro Schröder
a, Christian F. Doeller
a,c,daKavli Institute for Systems Neuroscience, NTNU – Norwegian University of Science and Technology, MH, NTNU, Postbox 8905, Trondheim 7491, Norway
bDepartment of Physics, NTNU – Norwegian University of Science and Technology, Trondheim, Norway
cMax Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany
dInstitute of Psychology, Leipzig University, Leipzig, Germany
a r t i c le i n f o
Keywords:
Magnetic resonance imaging Diffusion tensor imaging Structural connectivity Medial entorhinal cortex Lateral entorhinal cortex Segmentation
a b s t r a ct
Themedial(MEC)andlateralentorhinalcortex(LEC),widelystudiedinrodents,arewelldefinedandcharacter- ized.Inhumans,however,theexactlocationsoftheirhomologuesremainuncertain.Previousfunctionalmagnetic resonanceimaging(fMRI)studieshavesubdividedthehumanECintoposteromedial(pmEC)andanterolateral (alEC)parts,butuncertaintyremainsaboutthechoiceofimagingmodalityandseedregions,inparticularinlight ofasubstantialrevisionoftheclassicalmodelofECconnectivitybasedonnovelinsightsfromrodentanatomy.
Here,weusedstructural,notfunctionalimaging,namelydiffusiontensorimaging(DTI)andprobabilistictrac- tographytosegmentthehumanECbasedondifferentialconnectivitytootherbrainregionsknowntoproject selectivelytoMECorLEC.WedefinedMECasmorestronglyconnectedwithpresubiculumandretrosplenialcor- tex(RSC),andLECasmorestronglyconnectedwithdistalCA1andproximalsubiculum(dCA1pSub)andlateral orbitofrontalcortex(OFC).AlthoughourDTIsegmentationhadalargermedial-lateralcomponentthaninthe previousfMRIstudies,ourresultsshowthatthehumanMECandLEChomologueshaveaborderorientedboth towardstheposterior-anteriorandmedial-lateralaxes,supportingthedifferentiationbetweenpmECandalEC.
1. Introduction
The entorhinal cortex (EC) is a part of the medial temporal lobe, and a centralstructure for memory formation and navigation (Eichenbaumetal.,2007;MoserandMoser,2013;SuzukiandEichen- baum,2000).Itisclassically viewedasahubfor processingandre- layinginformationfrom theneocortextothehippocampus,andvice versa(Buzsáki,1996;LavenexandAmaral,2000).TheEC canbedi- videdintotwomainsubregions– ‘medial’entorhinalcortex(MEC)and
‘lateral’entorhinalcortex(LEC)– whichdifferinbothfunctionalproper- tiesandconnectivitywithotherregions(Cantoetal.,2008;Kerretal., 2007;vanStrienetal.,2009).Boththefunctionandanatomyofthe ECsubregionshavebeenwidelystudiedinrodentsandnon-humanpri- mates.Basedmainlyonresearchinrodents,theMECisassociatedwith spatialprocessinginaglobal,allocentricframeofreference,giventhe prevalenceof spatiallymodulatedcells suchasgrid andheaddirec- tioncells(Fyhnetal.,2004;Haftingetal.,2005;Høydaletal.,2019; Knierimetal.,2014).Incontrast,theLECcontainscellsthataresen- sitivetothepresenceofobjectsinalocalframeofreferenceorpro- cessingoftime(DeshmukhandKnierim,2011;Knierimetal., 2014; Tsaoetal.,2013,2018).Althoughrecentyearshaveseenastarkin-
∗Correspondingauthor.
E-mailaddress:ingrid.f.syversen@ntnu.no(I.F.Syversen).
creaseinfunctionalmagneticresonanceimaging(fMRI)studiesofthe humanEC(Bellmundetal.,2019;Chenetal.,2019;Montchaletal., 2019; Maass etal., 2015;Navarro Schröderetal., 2015; Reaghand Yassa,2014;Schultzetal.,2012),theexactlocationsofthehumanho- mologuesofMECandLECremainuncertain.
Whilecomprehensiveentorhinaldelineationsbasedoncytoarchitec- tonicanalysesexist(Insaustietal.,1995;Krimeretal.,1997),wecannot directlyrelatethesetodatasetsobtainedwithMRIwheretheresolution doesnotcoverthesingleneuronlevel.Moreover,alsointhemacaque monkey,thecytoarchitectonicallydefinedsubdivisionsofEChavenot yetyieldedacleardistinctionbetweenwhatmightbecounterpartsof MECandLECintherodent,anditwassuggestedthatconnectionaldata mightbeamorefruitfulapproach(WitterandAmaral,2021).Thislack ofcleardefiningcriterialimitstheinterpretationoffindingsinvolving ECinhumansassessedwiththehelpofMRimages.Forexample,anin- depthparcellationofthehomologueregionsofMECandLECinhumans ishighlyimportantforourunderstandingoftheroleoftheECinspatial (Bellmundetal.,2016;Doelleretal.,2010;Howardetal.,2014)and temporal(Bellmundetal.,2019; Montchaletal.,2019)contextrep- resentationsforepisodicmemoryandmnemonicbehavioringeneral.
Furthermore,results fromrodentmodelsofAlzheimer’sdiseaseindi- catethatthediseaseinitiallyaffectsLEC(Khanetal.,2014),whereas
https://doi.org/10.1016/j.neuroimage.2021.118723.
Received30July2021;Receivedinrevisedform22October2021;Accepted11November2021 Availableonline12November2021.
1053-8119/© 2021TheAuthors.PublishedbyElsevierInc.ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/)
studiesonhumansshowthatthediseaseaffectsanterolateralpartsof ECbefore moreintermediateorposteromedialpartsbecomeaffected (Berronetal.,2021;Kulasonetal.,2019).Formulatingcriteriaother thananatomicallandmarkstodefinethecounterpartofMECandLEC inhumanswillthereforebedirectlyrelevantfordesignandinterpreta- tionofstudiesontheirroleincognitionandfunctionaldeclineinrodent modelsinwaysthatallowforextrapolationtohumans,andviceversa.
fMRIstudieshaveindeedshownthatcertainpropertiesoftherodent andnon-humanprimateECalsoapplytothehumanEC(Doelleretal., 2010;ReaghandYassa,2014;Schultzetal.,2012).Basedonthesubdi- visionoftherodentECintoMECandLEC,studieshavetriedtolocalize theirrespectivehomologueregions inhumans.PreviousfMRIstudies testedconnectivity’fingerprints’ofECsubregionstootherpartsofthe brain.Studiesinrodentsandnon-humanprimateshavedemonstrated alargelysimilarorganizationofECconnectivityacrossspecies(Canto etal.,2008),thuspredictingdistinctfMRIconnectivityfingerprintsfor thetwosubregionsinhumansaswell.Theresultingdelineationsofpu- tativehumanhomologueregionsoftherodentMECandLECwerela- beledposteromedialEC(pmEC)andanterolateralEC(alEC),basedon theoutcomeoftwoindependentfMRIstudiesthattestedlocalandglobal connectivity,respectively(Maassetal.,2015;NavarroSchröderetal., 2015).However,itremainsunclearwhethertheresultswereaffectedby thenatureoftheimagingmodalityorthechoiceofseedbrainregions usedtoidentifythesubregions.
Inadditiontotheneuroimagingmodality,thesecondreasonfora re-evaluationhasgained additionalimportancesincetheassumption about EC connectivity on which parts of the previous fMRI studies (Maassetal.,2015)werebasedonhasbeenrecentlyrevised.Foryears, theexistenceoftwoparallelcorticalconnectivitystreamsthroughthe EChasbeentheacceptedmodel(Nilssenetal.,2019;Ranganathand Ritchey,2012;Witteretal.,2017).Thiscomprisesonepathwayintothe hippocampus viathe parahippocampal/postrhinal cortex(PHC/POR) andMEC(the“where” pathway),andaparallelpathwayviatheperirhi- nalcortex(PRC)andLEC(the“what” pathway).However,recentevi- dencesubstantiallychallengedthis view.Doanandcolleaguesfound thatPORinrats,whichcorrespondstothePHCinhumans,doesalso projecttoLEC.Theseauthorsfurtherarguethatexistingdatainmonkeys substantiatethisnotion(Doanetal.,2019).Thisisinlinewithnewfind- ingsinhumansindicatingthatthehippocampal-entorhinal-neocortical connectionsarefarmorecomplexthanapuresegregationinto“where” and“what” pathways(Huangetal.,2021).
InordertoidentifythehumanhomologuesofMEC andLEC, we shouldtakeadvantageofknownuniqueconnectionstoeachsubregion.
Forexample,inrodentsthepresubiculumprojectsalmostexclusively toMEC,whereasdistalCA1andproximalsubiculum(dCA1pSub,i.e.
theborderregionbetweenCA1andsubiculum)projectmoststrongly toLEC(Caballero-BledaandWitter,1993;HondaandIshizuka,2004; Witter and Amaral, 1991, 2021). Meanwhile, the retrosplenial cor- tex(RSC) andthelateral orbitofrontalcortex (OFC)arerespectively selectively connectedwithMEC andLEC(Hoover andVertes,2007; InsaustiandAmaral,2008;JonesandWitter,2007;Kondo andWit- ter,2014;Saleemetal.,2008;WitterandAmaral,2021;WyssandVan Groen,1992).Toinvestigatetheconnectivitybetween theseregions, thereareseveralimagingmodalitiesavailable.Analternativemethod tothewidelyusedfMRIandfunctionalconnectivityistostudyinstead structuralconnectivity usingdiffusiontensor imaging(DTI),another typeofMRI(Powelletal.,2004;Zeinehetal.,2012).Here,oneexploits thediffusionofwatermoleculesinsidewhitemattertractsandusesthis tomapthepathsofthesefibers– so-calledtractography(Morietal., 1999;MoriandZhang,2006).MappingDTIconnectivityprofilesfrom corticesthatprojectselectivelytoeitherECsubregioncouldprovidea novellineofevidencetoidentifyMECandLEC(Ezraetal.,2015;Máté etal.,2018;Sayginetal.,2011).
Theobjectiveofthisstudyisthereforetoidentifythehumanhomo- loguesoftherodentMECandLECusingDTI,incorporatingthenovel insightsfromrodentanatomy.Toachievethis,weperformedproba-
bilistictractographyonhigh-qualityDTIdataacquiredbytheHuman ConnectomeProject(Fanetal.,2016).WeidentifytheECsubregionsby analyzingtheconnectivityprofilesfromregionsofinterest(ROIs)that projectselectivelytoeitherofthemandcomparethesetotheresults frompreviousfMRIstudies.
2. Materialsandmethods 2.1. MRIdata
PubliclyavailablestructuralanddiffusionMRIdatafrom35healthy adultswereobtainedfromtheMGH-USCHumanConnectomeProject database (https://ida.loni.usc.edu, http://db.humanconnectome.org), inlinewiththeMGH-USCHCPDataAgreement.Allparticipantspro- vided writteninformed consent,andtheexperiments wereapproved by theinstitutional reviewboardof PartnersHealthcare (Fan etal., 2016). The data were acquired on a Siemens 3T Connectom scan- ner with maximum gradient strength of 300 mT/m and slew rate 200 T/m/s(McNabet al.,2013; Setsompop etal., 2013).Structural T1-weightedimageswereacquiredusinga3Dmagnetization-prepared rapid gradient-echo (MPRAGE) sequence at 1 mm isotropic resolu- tion.Diffusiondatawereacquiredusingaspin-echoecho-planarimag- ing (EPI)sequence at 1.5 mm isotropicresolution, withb-values of 1000s/mm2(64directions),3000s/mm2(64directions),5000s/mm2 (128directions)and10,000s/mm2(256directions).Onenon-diffusion- weighted(b=0)imagewascollectedevery14imagevolumes.
2.2. Preprocessing
TheMRIdatawereminimallypreprocessedbytheHumanConnec- tomeProjectasdescribedinFanetal.(2014).Inbrief,thispreprocessing pipelineincludedgradientnonlinearitycorrection,motioncorrection, Eddycurrentcorrectionandb-vectorcorrection.
2.2.1. Registration
Both structural and diffusion images were brain extracted us- ing the brain mask from running the FreeSurfer (version 7.1.1, https://surfer.nmr.mgh.harvard.edu/) functionsrecon-allanddt-recon on the participant’s structural and diffusion images, respectively (Fischletal.,2002,2004),beforerefiningtheresultusingtheFMRIB Software Library’s (FSL; version5.0.9, http://fsl.fmrib.ox.ac.uk/fsl/) functionBET(Jenkinsonetal.,2012;Smith,2002).Forthediffusion images,brainextractionandregistrationwereperformedonthepar- ticipant’saverageb=1000image.Theindividualbrain-extractedstruc- turalanddiffusionimageswereregisteredtoeachother,aswellastothe MNI152–09bstandardbraintemplate(Fonovetal.,2009),usingsym- metricnon-linearregistrationintheAdvancedNeuroimagingToolbox (ANTs;version2.3.4,http://stnava.github.io/ANTs/)basedonmutual information(Avantsetal.,2011).
2.2.2. Regionsofinterest
Regionsofinterest(ROIs)includingtheEC,presubiculum,CA1and subiculumwere extractedfrom theautomated corticalandsubcorti- cal parcellationobtainedfrom runningFreeSurfer’srecon-allandseg- mentHA_T1functionsontheMNI152-09btemplate(Fischletal.,2002, 2004;Iglesiasetal.,2015).TheECROIwasfurtherrefinedbymasking itbyaprobabilisticECROI,thresholdedat0.25fromtheJülich-Brain Cytoarchitectonic Atlas(Amuntsetal., 2020).Sincetheresulting EC ROIextendedtoofarposteriorlytowardstheparahippocampalcortex andlaterallybeyondthecollateralsulcus,wealsoperformedamanual adjustment.ThismanualadjustmentcomprisedusingtheFSLfunction fslmaths-erotoerodetheROIonce,beforeremovingremainingvoxels withanirregularappearanceinposteriorandlateralparts.Wethencre- atedROIsofdistalCA1/proximalsubiculumbysplittingeachofthetwo hippocampalstructuresinhalfalongitsproximodistalaxis.Ofallvox- elsencompassingCA1,thehalflocateddistallywasincluded,andofall
thevoxelsencompassingsubiculum,thehalflocatedproximallywasin- cluded:thesetwohalvesthusmakeupwhatweheredefineandreferto as‘distalCA1/proximalsubiculum’(dCA1pSub).TocreateRSCandOFC ROIs,respectively,theFreeSurferparcellationsnamed“isthmuscingu- late” and“lateralorbitofrontal” wereusedasastartingpoint.Thefinal RSCROIwasobtainedbytailoringtheisthmuscingulateandremoving theexcesssuperiorareas,whilethefinalOFCROIwasobtainedbyex- tractingtheposterolateralquadrantofthelateralorbitofrontalarea.All resultingROIsareshowninSupplementaryFig.1.TheROIswerereg- isteredtotheparticipants’individualspacesbyapplyingthecalculated transformationsfromANTs.Toincreasetheanatomicalprecisionofthe ROIs,theregisteredROIswerethenmaskedbyrespectiveparticipant- specificFreeSurferparcellations.
2.3. DTIanalysis
AllDTIanalyses wereperformedin theparticipant’snativediffu- sionspace.Voxel-wisefiberorientationdistributionfunctions(fODFs) werecomputedbyrunningtheFSLfunctionbedpostxonthediffusion data,using thezeppelindeconvolutionmodel,aRician noisemodel, andburn-inperiod3000(Sotiropoulosetal.,2016).Probabilistictrac- tographybetweentheECandpresubiculum,dCA1pSub,RSCandOFC ROIswasthenperformedbyrunningFSL’sprobtrackx2onthefODFs (Behrensetal.,2007,2003b).TractographywasperformedbothinROI- by-ROIandvoxel-by-ROIconnectivitymode,withnumberofsamples 250,000,minimalpathlength5mm,andamidlineterminationmask (Behrensetal.,2003a;Ezraetal.,2015;Johansen-Berg etal., 2004; Máté etal.,2018;Sayginetal.,2011).FortractographybetweenEC andpresubiculum,pathswereexcludediftheyreachedthedCA1pSub ROI,whilefortractographybetweenECanddCA1pSub,pathswereex- cluded iftheyreachedthepresubiculum ROI – andequivalently for tractographybetweenEC andRSC/OFC.Forbothbedpostxandprob- trackx2,parameterswererunwithdefaultvaluesunlessotherwisespec- ified.ROI-by-ROIconnectivitymodeprovidesprobabilitymapsofthe connectivitypathsbetweentheROIs,whilevoxel-by-ROIconnectivity modeprovidesprobabilitymapsofthevoxel-wiseconnectivityofthe ECROIwiththeotherROIs,respectively.Alltractographyresultswere registeredtoMNIspaceandfurtheranalyseswereperformedthereto facilitateinter-participantcomparisons.
2.4. MECandLECsegmentation
Thevoxel-wiseconnectivitymapswerenormalizedto[0,1]bydi- viding thembythemaximumprobability foreach hemisphere sepa- rately,andthenthresholdedby0.01toreducefalsepositiveconnections (Behrensetal.,2003a;Sayginetal.,2011).Thisthresholdwasdeter- minedempiricallybytestingarangeofthresholdsandchoosingtheone thatinmostcasesremovedconnectionsoutsidethegraymatter,because duetoremainingdistortionsintheDTIimagessomeof theEC ROIs slightlyextendedintoairvoxels.Crucially,wethendefinetheMECas theregionthatismoststronglyconnectedwiththepresubiculumand/or RSC,whiletheLECistheregionthatismoststronglyconnectedwith dCA1pSuband/orOFC(Caballero-BledaandWitter,1993;Hondaand Ishizuka,2004;HooverandVertes,2007;InsaustiandAmaral,2008; JonesandWitter,2007;KondoandWitter,2014;Saleemetal.,2008; WitterandAmaral,1991,2021;WyssandVanGroen,1992).Foreach participant,ahardsegmentationwasperformedonthenormalizedand thresholded voxel-wise connectivity maps using FSL’s find_the_biggest (Behrensetal.,2003a;Johansen-Bergetal.,2004),meaningthatthe voxels thathad ahigher connection probability withthe presubicu- lum/RSCthanwithdCA1pSub/OFCwereclassifiedasMEC,andvice versaforLEC.
2.5. Groupanalysis
GroupprobabilitymapsoftheconnectivitypathsbetweentheROIs, as well as group probability mapsof voxel-wise connectivity, were created by summing and averaging all the individual maps. Inter- participantsegmentationvariabilitymapswerecreatedbyaddingto- getheralltheindividualparticipants’MECandLECsegmentations,re- spectively.GroupMECandLECsegmentationwereperformedsimilarly totheindividualsegmentation:Thegroupvoxel-wiseconnectivitymaps werefirstsmoothedwithaGaussiankernelof1mmandthresholdedby 0.01,andthenahardsegmentationwasperformedequivalentlytothe single-participantsegmentationbycomparingtheconnectionprobabil- itiesofECwithpresubiculum/RSCvs.dCA1pSub/OFC.Fourdifferent segmentationswereperformedwithallthe2×2combinationsofseed regions,inadditiontoacombinedsegmentationapproachwherethe connectivitymapsforpresubiculum+RSCandfordCA1pSub+OFC, respectively,werecombinedandaveragedbeforesegmentation.
2.6. Segmentationcomparisons
Toassessthedifferentsegmentationapproachesandcomparethere- sultinglocationsofMECandLEC,wecalculatedtheorientationofthe MEC-LECborderalongtheposterior-anterior (PA)andmedial-lateral (ML)axes,respectively.Thiswasperformedbyfirstcalculatingthecen- tersofgravityofthedifferentlydefinedMECsandLECs,andthevector betweenthesecentersofgravity.Next,theanglebetweenthisvector andapurePAorMLvectorwasdetermined.WedefinedthePAaxisas thelongaxisofthehippocampus.ThedegreeofPA-orML-orientedbor- derwasthendefinedbetween0and100%suchthatanangleof0° tothe PAorMLvectormeansthattheborderis100%orientedalongthePAor MLvector,respectively.Correspondingly,anangleof90° wouldmean thattheborderis0%orientedalongtherespectiveaxis,i.e.itisorthogo- naltothataxis.Inaddition,thedifferentsegmentationswerecompared withrespecttothesizesoftheresultingMECsandLECs,andthesizera- tiosbetweenthesewerecalculated.Allthesesegmentationcomparisons werealsocarriedoutonthetwofMRI-basedsegmentationsofpmEC andalECavailablefordownloadfromearlierstudies(Maassetal.,2015; NavarroSchröderetal.,2015).
3. Results
ToqualitativelyvisualizetheconnectivitypathsbetweentheECand theregionshypothesizedtobeconnectedwithitssubregions,weran probabilistictractographybetweentheregions.Byseedingpathsfrom allvoxelsintheEC,presubiculum,dCA1pSub,RSCandOFCROIs,maps oftheconnectivitypathsbetweentheECandtheotherROIswerecre- ated. Theresultinggroupaveraged pathsareshown inFig.1.Inall figures,bluecolorschemesareusedforMEC-relatedregions,i.e.pre- subiculumandRSC,whileredcolorschemesareusedforLEC-related regions,i.e.dCA1pSubandOFC.Themapsshowthatalltheregionsex- hibitclearconnectivitywiththeEC.ConnectionswithdCA1pSubextend furtheranteriorlyintheECthantheconnectionswiththepresubiculum, andtheconnectionswithpresubiculumandRSCseemtotakeasimilar routetotheEC.ThepathsbetweenOFCandEC,however,standout fromtheothersastheytakeamorelateralroute,buttheinferiorpart seemstopassclosetodCA1pSub.Notethatthecolormapintensityin thesemapsdoesnotrepresenttheactualnumberofwhitemattertracts, butinsteadscaleswiththeprobabilitythatthetruepathbetweenthe ROIsliesinthatpoint.Correspondingconnectivitypathsforoneexam- pleparticipantareshowninSupplementaryFig.2.
BecausewewantedtosegmenttheECintotheMECandLEChomo- loguesbasedontheconnectivitywithotherregions,avoxel-by-voxel measureofconnectivityprobabilitywasneeded.Wethereforealsoran thetractographyonlyseedingfromtheECROIs.Then,foreachvoxelin theROI,wecountedhowmanyoftheseededpathsreachedtheother
Fig.1. GroupaverageconnectivitypathsbetweenECandpresubiculum,dCA1pSub,RSCandOFC.Connectivitypatternsareshownonsagittal(left)andcoronal (right)slicesinMNIspace,with“R” denotingtherightsideofthebrain.Thecolormapintensityrepresentsthenumberofprobabilisticpathsrunningthroughthat voxel.A:PathsbetweenECandpresubiculum,B:PathsbetweenECanddCA1pSub,C:PathsbetweenECandRSC,D:PathsbetweenECandOFC.(Forinterpretation ofthereferencestocolorinthisfigure,thereaderisreferredtothewebversionofthisarticle.)
ROIs.Theseconnectivitycountswerenormalizedtoaprobability,pro- vidingconnectivitymapsfortheECwiththeotherfourROIs.There- sultingsmoothedandthresholdedgroupaveragedconnectivitymapsare showninFig.2.Thesagittalslicesshowthattheconnectivitywithpre- subiculumandRSCappearstobestrongestintheposteriorpartofthe EC,whereastheconnectivitywithdCA1pSubandOFCisstrongestante- riorlyintheEC.Further,thepresubiculumconnectivitydoesnotshowa
clearmedial-lateralgradient,buttheconnectionswithdCA1pSub,RSC andOFCarestrongerlaterallyintheECintheselectedcoronalslices.
Correspondingconnectivitymapsforoneexampleparticipantareshown inSupplementaryFig.3.
ForsegmentationintotheMECandLEChomologues,themainhy- pothesiswas thattheseregions couldbeidentifiedbasedonconnec- tivitywithpresubiculumvs. dCA1pSub,respectively.Theactual seg-
Fig.2. GroupaveragemapsofECconnectivitywithpresubiculum,dCA1pSub,RSCandOFC.Themapsareshownonsagittal(left)andcoronal(right)slicesin MNIspace,with“R” denotingtherightsideofthebrain.ThecolormapintensityrepresentsthefractionofpathsseededfromthatECvoxelthatreachedtheother ROI.A:ECconnectivitywithpresubiculum,B:ECconnectivitywithdCA1pSub,C:ECconnectivitywithRSC,D:ECconnectivitywithOFC.(Forinterpretationofthe referencestocolorinthisfigure,thereaderisreferredtothewebversionofthisarticle.)
mentationwasperformedonavoxel-by-voxellevelintheECdetermin- ingwithwhichoftheothertworegionstheconnectionprobabilitywas highest,usingtheconnectivity mapsdescribedin thepreviouspara- graph.Forcomparison,theMEC-LECsegmentationwasalsoperformed basedonconnectivitywithRSCvs.OFC,respectively.Thiswasfirstper- formedindividuallyforallparticipants,andinter-participantsegmenta- tionvariabilitymapsforthepresubiculumvs.dCA1pSubandRSCvs.
OFCsegmentationapproachesareshown inFig.3.Formost partici- pants,MECisclearlylocatedmoreposteriorlyandLECislocatedmore
anteriorlyforbothsegmentationapproaches,andinadditiontheyare locatedmoremediallyandlaterallywithrespecttoeachotherforthe presubiculumvs.dCA1pSubapproach.TheRSCvs.OFCapproachalso showsthismedial-lateraltrendofMECandLECacrossparticipants,al- thoughnotasclearasforpresubiculumvs.dCA1pSub.Corresponding MECandLECsegmentationsforoneexampleparticipantareshownin SupplementaryFig.4.
Thesameconnectivity-basedMEC-LECsegmentationwasperformed onagrouplevelusingthegroupaveragedconnectivitymapsfromFig.2.
Fig.3. Inter-participantsegmentationvariabilitymapsfordifferentsegmentationapproaches.Resultsareshownonsagittal(left)andcoronal(right)slicesinMNI space,with“R” denotingtherightsideofthebrain.ThecolormapintensityrepresentsthenumberofparticipantsforwhichthatvoxelwasclassifiedasMECorLEC, respectively.A:MECpredictionbasedonhigherconnectivitywithpresubiculumthanwithdCA1pSub,B:LECpredictionbasedhigherconnectivitywithdCA1pSub thanwithpresubiculum,C:MECpredictionbasedonhigherconnectivitywithRSCthanwithOFC,D:LECpredictionbasedonhigherconnectivitywithOFCthan withRSC.(Forinterpretationofthereferencestocolorinthisfigure,thereaderisreferredtothewebversionofthisarticle.)
Asdescribed above,thegroup segmentationwas alsoperformedus- ingtwodifferentapproaches– presubiculumvs.dCA1pSub,andRSC vs.OFC– andtheresultingsegmentationsareshowninFig.4.Wesee thatfortheMECandLECpredictionsfrompresubiculumvs.dCA1pSub, thereisaclearmedial-lateral(ML)andposterior-anterior(PA)-oriented border between thesubregions. ForRSC vs. OFC, however, thePA- orientedborderismostprominent,butitisalsoslightlyML-oriented,
mostvisibleintheleftEC.Becausetheresultsfromthetwoapproaches wereslightlydifferent,wealsotriedtointerchangetheROIcombina- tions, andMECandLECsegmentationsfromusing presubiculumvs.
OFCandRSCvs.dCA1pSubcanbeseeninSupplementaryFig.5.Fur- thermore,toincludealltheinformationfromthe2×2combinations ofseedregionsintoonefinalsegmentation,weperformedanotherap- proachwhereweaveragedtheconnectivitymapsforpresubiculumand
Fig.4. GroupsegmentationsofMECandLECfromdifferentapproaches.Resultsareshownonsagittal(topleft)andcoronal(topright)slicesand3D-rendered(bottom left)inMNIspace.TheMECandLECpredictionsareshowninblueandred,respectively.A:MECandLECpredictionbasedonconnectivitywithpresubiculum vs.dCA1pSub,B:MECandLECpredictionbasedonconnectivitywithRSCvs.OFC.S=superior,I=inferior,A=anterior,P=posterior,R=right,L=left.(For interpretationofthereferencestocolorinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.).
Table1
Degreeofposterior-anterior(PA)ormedial-lateral(ML)-orientationoftheborderbetweenMECandLECfor differentsegmentationapproaches.ThedegreeofPA-orML-orientationisgivenasapercentagebetween0and 100%,dependentontheanglebetweentheMEC-LECcenterofgravityvectorandthepurePAorMLvector, respectively.Allnumbersaregivenasthemeanofbothhemispheres±meanabsolutedeviation.
Segmentation approach
Posterior-anterior (PA) axis Medial-lateral (ML) axis Angle (°) % PA Angle (°) % ML DTI Presubiculum/dCA1pSub 45.1 ± 3.0 49.9 ± 3.3 29.8 ± 1.9 66.9 ± 2.1
RSC/OFC 39.7 ± 15.3 55.9 ± 17.0 84.5 ± 11.1 6.1 ± 12.3 Presubiculum/OFC 41.3 ± 17.3 54.1 ± 19.3 81.2 ± 10.3 9.8 ± 11.5 RSC/dCA1pSub 33.9 ± 1.2 62.3 ± 1.3 64.9 ± 18.8 27.9 ± 20.9 Presubiculum + RSC/dCA1pSub + OFC 37.5 ± 10.6 58.3 ± 11.8 73.1 ± 17.4 18.7 ± 19.3 fMRI Navarro Schröder et al. 6.8 ± 1.8 92.4 ± 2.0 85.5 ± 0.7 5.0 ± 0.8
Maass et al. 6.6 ± 0.5 92.7 ± 0.6 87.9 ± 0.9 2.4 ± 1.0
Table2
ResultingsizesofMECandLECfordifferentsegmentationapproaches,andthesizeratiobetween MECandLEC.ThenumbersofvoxelsaregivenfortheROIsinMNIspacewith0.5mmisotropic resolution.
Segmentation approach
Size (# voxels)
MEC/LEC size ratio
MEC LEC
DTI Presubiculum/dCA1pSub 12,759 7763 1.64
RSC/OFC 12,971 8727 1.49
Presubiculum/OFC 13,614 6979 1.95
RSC/dCA1pSub 11,045 10,282 1.07
Presubiculum + RSC/dCA1pSub + OFC 13,571 7379 1.84 fMRI Navarro Schröder et al. 12,802 16,028 0.80
Maass et al. 3776 11,008 0.34
RSC,andthemapsfordCA1pSub andOFC(Fig.5AandB).Fig.5C showstheresultingMECandLEChomologuesfromthiscombinedseg- mentationapproach.Withthisapproach,aswithseparatecombinations ofseedregions,wefindbothaPA-andML-oriented(althoughmostvis- ibleinthelefthemisphere)borderbetweenMECandLEC.Thesefinal MECandLECmasksarealsoavailableintheSupplementaryfiles.
Inanextstep,sincethebordersofthesegmentationsfromdifferent approachesshowedslightlydifferentorientationsalongtheposterior- anterior (PA) and medial-lateral (ML) axes, we wanted to quantify thisdirectionaldifferencebycalculatingthe“degree” ofPA-andML- orientationoftheborders.Thiswasdefinedasapercentagefrom0to 100%,dependentontheanglebetweentheMEC-LECcenterofgravity vectorandapurePAorMLvector.Table1showstheresultingdegreesof PA-vs.ML-orientedbordersforthedifferentsegmentationapproaches includingthefMRIsegmentationsfrompreviousstudies(Maassetal., 2015;NavarroSchröderetal.,2015).Thecenterofgravityvectorsare alsoplottedinacommonreferenceframeinSupplementaryFig.6.All DTIsegmentationapproacheshaveaborderwithaPA-orientationof around50–60%,andavaryingdegreeofML-orientationfrom6%for RSCvs.OFCupto67%forpresubiculumvs.dCA1pSub.Theborders betweenthesegmentationsfromfMRI haveahighPA-orientationof around92%,andalowerdegreeofML-orientationthanalloftheDTI approaches.Interestingly,whencomparingthedifferentcombinations ofDTIapproaches,usingdCA1pSubasthedefiningregionforLECyields ahigherdegreeofML-orientationthanusingOFC.Similarly,usingRSC asthedefiningregionforMECyieldsaslightlyhigherdegreeofPA- orientationoftheborderthanusingpresubiculum,butthisislesspromi- nent.
Finally,wewantedtocomparetheresultingsizesoftheMECand LEChomologuesfromallthedifferentsegmentationapproaches,and theseareshowninTable2.ForallDTIapproaches,theMECislarger thanLEC,whilefMRIontheotherhandyieldsalargerLECthanMEC.
ThesubregionsaremostequallysizedwhenusingtheRSCvs.dCA1pSub approach.
4. Discussion
In this study, we used DTI andprobabilistic tractography in 35 healthyadultstosegmentthehumanECintohomologuesofwhatin othermammalshavebeenfunctionally,connectionally,andcytoarchi- tectonicallydefinedasMECandLEC.Webasedthesegmentationon ECconnectivitywithfourbrainregionsknowntoselectivelyprojectto eitheroftheECsubregionsinmultiplespecies.Differentcombinations of thesefourregions allshowedbothaposterior-anterior(PA)anda medial-lateral(ML)-orientedborderbetweenthehumanhomologuesof MECandLEC.Thisorientationofthethusdefinedborderissimilarto thatdefinedinpreviousfMRIstudiesresultinginthedefinitionofthe twosubregionsaspmECandalEC(Maassetal.,2015;NavarroSchröder etal.,2015).NotehoweverthatourDTIresultsshowalargerdegreeof ML-orientation,andacorrespondinglylowerdegreeofPA-orientation oftheborderbetweenthesubregionscomparedtothepreviousfMRI results.
TheresultsfromourstudysubstantiatethepmECandalECsubdivi- sionofthehumanECsuggestedinpreviousfMRIstudies(Maassetal., 2015;NavarroSchröderetal.,2015).AlthoughsomeearlierfMRIstud- iesonmnemonic processingintheECfoundadissociationprimarily along themedial-lateralaxis(ReaghandYassa,2014;Schultzetal., 2012),itisimportanttorealizethateventheorientationofthecytoar- chitectonicallydefinedborderbetweenMECandLECinrodentsdoes not alignalong apuremedial-to-lateralaxis.Rather,theMECinro- dentsislocatedintheposterior-medialEC,andtheLECislocatedin theanterior-lateralEC(vanStrienetal.,2009).Also,inmacaquemon- keys,tracingstudiesshowdifferentialconnectivityincaudalvs.rostral portions(WitterandAmaral,2021).Apuremedial-lateralsubdivision of humanECis thusnottobeexpected.Nevertheless,thesomewhat differentorientationsoftheborderbetweenthehumanhomologuesof MECvs.LECsubdivisionsfoundusingDTIvs.fMRIstudiesraisesthe questionofwhichofthetwoimagingmodalitiesshouldbepreferredto definethepositionandorientationofthisborder.
Fig.5. Groupconnectivitymapsandsegmentationusingacombinedapproachwithpresubiculum+RSCvs.dCA1pSub+OFC.A:ECconnectivitywithpre- subiculum+RSCcombined.B:ECconnectivitywithdCA1pSub+OFCcombined.C:MECandLECpredictionbasedonconnectivitywithpresubiculum+RSCvs.
dCA1pSub+OFCcombined.S=superior,I=inferior,A=anterior,P=posterior,R=right,L=left.(Forinterpretationofthereferencestocolorinthisfigure,the readerisreferredtothewebversionofthisarticle.)
There areseveral possible explanationsasto whyourDTI study showedslightlydifferent segmentationresults thanthefMRIstudies.
First,DTIandfMRIaretwodifferentimagingmodalitieswithinherently differentmechanismsofconnectivity.WhileDTIexploitsthediffusion ofwatermoleculesinordertotracethestructuralpathsofconnectiv- itybetweenbrainregions(Morietal.,1999; MoriandZhang,2006; Powelletal.,2004;Zeinehetal.,2012),fMRIidentifiesfunctionalcon- nectivitybycorrelatingblood-oxygen-level-dependent(BOLD)signals acrosstime(VanDijketal.,2010).Althoughstructuralandfunctional connectivityintheoryshouldbecloselylinked,theyareinrealityquan- titativelydifficulttocomparebecauseofthecomplexityoftheconnec- tivitymechanismsof thebrain(HuangandDing,2016;Messé et al., 2015).Anotherreasonforthedifferentresultsbetweenthecurrentand thepreviousstudiescouldbetheuseofdifferentseedregionstoidentify theMECandLEChomologues.WhileweusedpresubiculumandRSCto defineMEC,anddCA1pSubandOFCtodefineLEC(Caballero-Bledaand Witter,1993; Hondaand Ishizuka, 2004; Hoover andVertes, 2007; JonesandWitter,2007;KondoandWitter,2014;Saleemetal.,2008; WitterandAmaral,1991,2021;WyssandVanGroen, 1992),oneof thefMRIstudiesinvestigateddifferentialconnectivityofPHCvs.PRC anddistalvs. proximal subiculum(Maass etal., 2015),whereas the otherusedregionsinaposterior-medialvs.ananterior-temporalcorti- calsystem(NavarroSchröderetal.,2015).Thenewinsightsfromrodent anatomyindicatethatwhilePHCareaTHisconnectedwiththeMEC, PHCareaTFisconnectedwiththeLEC(WitterandAmaral,2021).As areaTFislocatedmorelaterallythanTH,thismightinpartexplainwhy thepreviousfMRIstudywheretheyusedconnectivitywiththewhole PHCtodefinethepmEC(Maassetal.,2015)showedalowermedial- lateralcomponentoftheirpmEC-alECsegmentationthanourresults.In ordertodeterminetowhichextenteachofthesereasonscontributedto thedifferentsubdivisionresultsacrossstudies,bothimagingmodalities withidenticalseedregionsshouldbeinvestigatedandcomparedmore rigorouslyinonesingle,largercohortofparticipants.
Interestingly,usingdifferentseedregionstoidentifyMECandLEC resultedinvaryingdegreesofPA-andML-orientationoftheborderbe- tweenthem.ItisunclearwhetherthisisinherentlylinkedtotheDTI method,orduetoanactualconnectivitydifferencebetweentheregions.
UsingpresubiculumanddCA1pSubastheseedregions,whicharesit- uatedmediallyandlaterallywithrespecttoeachother,respectively, resultedinaborderwithhigherdegreeofML-thanPA-orientation.On theotherhand,usingRSCandOFC,whicharesituatedposteriorlyand anteriorlyinthebrain,respectively,resultedinaborderwithhigherde- greeofPA-thanML-orientation.Althoughitisnotunnaturaltoassume thatthebrainisorganizedsuchthatconnectedregionsaresituatedmore closelytoeachother,thiscouldalsobeaneffectofusingprobabilis- tictractography,wheretheapparentconnectivityprobabilitydepends one.g.thelengthofthepathandthesizeoftheROIs(Behrensetal., 2007).Inotherspecies,includingrodentsandmonkeys,thepresubicu- lumandRSCshowinputstotheECwithasimilarspatialdistribution (WitterandAmaral,2021),aligningwithourmapsofconnectivitypaths withthesetwoseedregions.However,comparingthedifferentMECand LECsegmentationsfromthedifferentseedregioncombinationsshows thatwhileinterchangingpresubiculumandRSCyieldsonlyslightlydif- ferentorientationoftheborderalongthePAandMLaxes,thedifference wheninterchangingdCA1pSubandOFCismoresubstantial.Inother species,dCA1pSubareknowntoprojecttobothrostralanddorsolat- eralpartsofEC,whereasposterolateralOFCmainlyprojectsdorsolat- erallyintheEC(InsaustiandAmaral,2008;KondoandWitter,2014; Saleemetal.,2008;WitterandAmaral,1991,2021).Whetherthesere- gionsinhumansprojecttodifferentpartsofthehomologueofLEC,or whetherourresultsareaffectedbyusingDTIandprobabilistictractog- raphy,shouldbefurtherinvestigatedbyalsocomparingECfunctional connectivitytotheseareasusingfMRI.Notealsothatthetopography ofprojectionsfromdCA1pSubalongthemedial-lateralaxisoftheEC dependsonwheretheseedisplacedalongtheposterior-anterioraxis ofthedCA1pSub(WitterandAmaral,2021),whichemphasizestheim-
portanceofcarefullydefinedseedROIs.Inaddition,evenmoreareas hypothesizedtobepreferentiallyconnectedtoMECorLECshouldbe evaluatedinthefuture.
Inordertodetermineandcomparetheconnectivitiesbetweenthe ECandtheotherROIs,wenormalizedtheconnectivitymapsbydivid- ingthembythemaximumprobabilityofeachmap.Thiscouldintro- duceabiasintheresults.Bydoingthis,weintrinsicallyassumethat themaximumconnectivitystrengthtoeachoftheotherROIsareequal, andthesegmentationprocessdoesnottakeintoaccountthattheMEC connectionsmightbestrongerthantheLECconnections,orviceversa.
However,littleisknownaboutthestrengthofconnectivitiesatthislevel ofdetail,particularlysinceitisnotstraightforwardtoexamineoreven defineconnectivitystrength.Connectivitystrengthsurelydependson axonaldensity,butotherfactorslikesynapticdensityandefficacyare otherimportantvariables.Nevertheless,evenifweweretoknowthat someoftheconnectionsarestrongerthantheothers,probabilistictrac- tographyprovidesarelativeinsteadofanabsolutemeasureofconnec- tivityandisalsodependentonpathlengths,ROIsizesandthenumberof possiblepathdirectionsinavoxel.Normalizingtheconnectivitymaps basedondifferentconnectivitystrengthswouldthereforebeahighly complex task.Therefore,wedidnot imposeany furtherassumptions aboutconnectivitystrengthsinouranalyses.
Performing tractographybetween theECandthefourotherROIs showedclearconnectivitypathsbetweentheareas.Thepurposeofthis analysiswastoverifythatthehypothesizedseedregionswereindeed structurallyconnectedwiththeEC.However,characterizationofthetra- jectoriesofthesepaths,includingtheirdifferentialdistributionwithin thewhitematteroftheangularbundle,wasoutsidethescopeofour study. Althoughqualitativelysimilar probabilisticpaths betweenthe ECandthesubicularcorticeshavebeenshowninpreviousexvivoDTI data(Augustinacketal.,2010),thelargedifferenceinspatialresolution betweenthestudiesmakesadirectcomparisondifficult.Thisisnever- thelessaninterestingtopicthatcouldbeaddressedinfuturestudies.
Ourstudyhassomelimitations.TodefineourROIs,wechosetouse regionsfromautomaticcorticalsegmentationprotocols.Thiscouldhave influencedtheanatomicalprecisionofouranalysis.Manualsegmenta- tionwouldbelabor-intensiveandrequireshighskillsinneuroanatomy, possiblylimitingthenumberofparticipantsthatcouldbeincludedin thestudy.However,wemanuallyadjustedsomeoftheautomatically segmented ROIs, andalsointersected theregistered ROIs from MNI spacewiththeparticipants’individualautomaticsegmentationsinorder toincreasetheanatomicalaccuracy.Anotherlimitationisthatthereare inherentchallengestotheEPIsequenceusedfordiffusionimaging.This resultsinagenerallylowsignal-to-noiseratiointheECandthewhole medialtemporallobe.Inaddition,theseregionsappeargeometrically distortedintheEPIimages,andalthoughthishasbeencorrectedfor,it isnotpossibletorecoverallofthelostsignal.Imperfectcorrectioncan alsoaffecttheaccuracyoftheROIs.Becauseoftheprobabilisticnature ofthetractographytechniqueitisunlikelythatnoisewillintroducefalse significantconnections,butitcanleavesomeconnectionsundetected.
Atlast,arelativelylownumberofparticipantswereincludedinour study,whichmighthaveinfluencedthestatisticalpoweroftheresults.
5. Conclusions
OurDTIresultssupportthedefinitionofpmECandalECashuman homologuesofMECandLECastheyarecurrentlyconnectionallyde- finedinrodentsandtoasomewhatlesserextentinmonkeys.Inspired bynovelinsightscomingfromrodentanatomy,wepresentasegmen- tation basedon acombination ofdifferentialpresubiculum/RSC and dCA1pSub/lateralOFCstructuralconnectivitywhichindicatesaborder betweenthetwosubdivisionsofECwithanorientationthatisangled bothtowardstheposterior-anterioraxis,aswellastothemedial-lateral axis.Thefactthattherearesomedifferencesintheorientationofthe borderbasedonDTIandfMRIdatainadditiontotheseedregionsused,
indicatestheneedforinvestigationinalargernumberofparticipants acrossbothmodalities.
Dataandcodeavailabilitystatement
The data used in this study were obtained from the MGH- USC Human Connectome project, and are publicly available from https://ida.loni.usc.eduandhttp://db.humanconnectome.org.
The following, freely available code/toolboxes were used in this study: FreeSurfer (version 7.1.1, https://surfer.nmr.mgh.harvard.edu/),FMRIBSoftwareLibrary’s(FSL, version5.0.9;http://fsl.fmrib.ox.ac.uk/fsl/),andAdvancedNeuroimag- ingToolbox(ANTs,version2.3.4;http://stnava.github.io/ANTs/).
CRediTauthorshipcontributionstatement
Ingrid Framås Syversen: Conceptualization, Methodology, Soft- ware,Formalanalysis,Investigation,Writing– originaldraft,Writing– review&editing,Visualization.MennoP.Witter:Methodology,Vali- dation,Writing– review&editing.AsgeirKobro-Flatmoen:Validation, Writing– review&editing.PålErikGoa:Conceptualization,Method- ology, Resources, Writing – review & editing, Supervision, Funding acquisition.TobiasNavarroSchröder:Conceptualization,Methodol- ogy,Validation,Writing– review&editing,Supervision.ChristianF.
Doeller:Conceptualization,Validation,Writing– review&editing,Su- pervision,Fundingacquisition.
DeclarationofCompetingInterest
Theauthorsdeclarenocompetinginterests.
Creditauthorshipcontributionstatement
IngridFramås Syversen: Conceptualization, Methodology, Soft- ware,Formalanalysis,Investigation,Writing– originaldraft,Writing– review&editing,Visualization.MennoP.Witter:Methodology,Vali- dation,Writing– review&editing.AsgeirKobro-Flatmoen:Validation, Writing– review&editing.PålErikGoa:Conceptualization,Method- ology, Resources, Writing – review & editing, Supervision, Funding acquisition.TobiasNavarroSchröder:Conceptualization,Methodol- ogy,Validation,Writing– review&editing,Supervision.ChristianF.
Doeller:Conceptualization,Validation,Writing– review&editing,Su- pervision,Fundingacquisition.
Acknowledgments
ThisstudywassupportedbytheNorwegianUniversityofScience and Technology (IFS: RSO-grant from the Faculty of Medicine and HealthSciences,projectnumber81850040).CFDisfurthersupported bytheMaxPlanckSociety,theEuropeanResearchCouncil(ERC-CoG GEOCOG724836), theKavliFoundation,theJebsenFoundation,the centerofExcellenceschemeoftheResearchCouncilofNorway– center forNeuralComputation(223262/F50),TheEgilandPaulineBraathen andFredKavlicenterforCorticalMicrocircuits,andtheNationalIn- frastructureschemeoftheResearchCouncilofNorway– NORBRAIN (197467/F50).
DatawereprovidedbytheHumanConnectomeProject,MGH-USC Consortium(PrincipalInvestigators: BruceR.Rosen,ArthurW. Toga andVanWedeen;U01MH093765)fundedbytheNIHBlueprintInitia- tiveforNeuroscienceResearchgrant;theNationalInstitutesofHealth grant P41EB015896; and theInstrumentation Grants S10RR023043, 1S10RR023401,1S10RR019307.
Supplementarymaterials
Supplementarymaterialassociatedwiththisarticlecanbefound,in theonlineversion,atdoi:10.1016/j.neuroimage.2021.118723.
References
Amunts, K., Mohlberg, H., Bludau, S., Zilles, K., 2020. Julich-Brain: a 3D probabilistic atlas of the human brain’s cytoarchitecture. Science 369, 988–992. doi: 10.1126/sci- ence.abb4588 .
Augustinack, J., Helmer, K., Huber, K., Kakunoori, S., Zöllei, L., Fischl, B., 2010. Direct visualization of the perforant pathway in the human brain with ex vivo diffusion tensor imaging. Front. Hum. Neurosci. 4. doi: 10.3389/fnhum.2010.00042 .
Avants, B.B., Tustison, N.J., Song, G., Cook, P.A., Klein, A., Gee, J.C., 2011. A reproducible evaluation of ANTs similarity metric performance in brain image registration. Neu- roimage 54, 2033–2044. doi: 10.1016/j.neuroimage.2010.09.025 .
Behrens, T.E., Johansen-Berg, H., Jbabdi, S., Rushworth, M.F., Woolrich, M.W., 2007.
Probabilistic diffusion tractography with multiple fibre orientations: what can we gain? Neuroimage 34, 144–155. doi: 10.1016/j.neuroimage.2006.09.018 . Behrens, T.E., Johansen-Berg, H., Woolrich, M.W., Smith, S.M., Wheeler-Kingshott, C.A.,
Boulby, P.A., Barker, G.J., Sillery, E.L., Sheehan, K., Ciccarelli, O., Thompson, A.J., Brady, J.M., Matthews, P.M., 2003a. Non-invasive mapping of connections between human thalamus and cortex using diffusion imaging. Nat. Neurosci. 6, 750–757.
doi: 10.1038/nn1075 .
Behrens, T.E., Woolrich, M.W., Jenkinson, M., Johansen-Berg, H., Nunes, R.G., Clare, S., Matthews, P.M., Brady, J.M., Smith, S.M., 2003b. Characterization and propagation of uncertainty in diffusion-weighted MR imaging. Magn Reson Med 50, 1077–1088.
doi: 10.1002/mrm.10609 .
Bellmund, J.L., Deuker, L., Doeller, C.F., 2019. Mapping sequence structure in the human lateral entorhinal cortex. eLife 8. doi: 10.7554/eLife.45333 .
Bellmund, J.L.S., Deuker, L., Navarro Schröder, T., Doeller, C.F., 2016. Grid-cell represen- tations in mental simulation. eLife 5, e17089. doi: 10.7554/eLife.17089 .
Berron, D., Vogel, J.W., Insel, P.S., Pereira, J.B., Xie, L., Wisse, L.E.M., Yushkevich, P.A., Palmqvist, S., Mattsson-Carlgren, N., Stomrud, E., Smith, R., Strandberg, O., Hans- son, O., 2021. Early stages of tau pathology and its associations with functional con- nectivity, atrophy and memory. Brain doi: 10.1093/brain/awab114 .
Buzsáki, G., 1996. The hippocampo-neocortical dialogue. Cereb. Cortex 6, 81–92.
doi: 10.1093/cercor/6.2.81 .
Caballero-Bleda, M., Witter, M.P., 1993. Regional and laminar organization of projections from the presubiculum and parasubiculum to the entorhinal cortex: an anterograde tracing study in the rat. J. Comp. Neurol. 328, 115–129. doi: 10.1002/cne.903280109 . Canto, C.B., Wouterlood, F.G., Witter, M.P., 2008. What does the anatomical organization of the entorhinal cortex tell us? Neural Plast. 2008, 381243.
doi: 10.1155/2008/381243 .
Chen, X., Vieweg, P., Wolbers, T., 2019. Computing distance information from landmarks and self-motion cues - differential contributions of anterior-lateral vs. posterior-medial entorhinal cortex in humans. Neuroimage 202, 116074.
doi: 10.1016/j.neuroimage.2019.116074 .
Deshmukh, S.S., Knierim, J.J., 2011. Representation of non-spatial and spatial informa- tion in the lateral entorhinal cortex. Front. Behav. Neurosci. 5, 69. doi: 10.3389/fn- beh.2011.00069 .
Doan, T.P., Lagartos-Donate, M.J., Nilssen, E.S., Ohara, S., Witter, M.P., 2019. Con- vergent projections from perirhinal and postrhinal cortices suggest a multisen- sory nature of lateral, but not medial, entorhinal cortex. Cell Rep. 29, 617–627.
doi: 10.1016/j.celrep.2019.09.005 , .e617 .
Doeller, C.F., Barry, C., Burgess, N., 2010. Evidence for grid cells in a human memory network. Nature 463, 657–661. doi: 10.1038/nature08704 .
Eichenbaum, H., Yonelinas, A.P., Ranganath, C., 2007. The medial temporal lobe and recognition memory. Annu. Rev. Neurosci. 30, 123–152. doi: 10.1146/an- nurev.neuro.30.051606.094328 .
Ezra, M., Faull, O.K., Jbabdi, S., Pattinson, K.T., 2015. Connectivity-based segmentation of the periaqueductal gray matter in human with brainstem optimized diffusion MRI.
Hum. Brain Mapp. 36, 3459–3471. doi: 10.1002/hbm.22855 .
Fan, Q., Nummenmaa, A., Witzel, T., Zanzonico, R., Keil, B., Cauley, S., Polimeni, J.R., Tisdall, D., Van Dijk, K.R., Buckner, R.L., Wedeen, V.J., Rosen, B.R., Wald, L.L., 2014.
Investigating the capability to resolve complex white matter structures with high b - value diffusion magnetic resonance imaging on the MGH-USC Connectom scanner.
Brain Connect. 4, 718–726. doi: 10.1089/brain.2014.0305 .
Fan, Q., Witzel, T., Nummenmaa, A., Van Dijk, K.R.A., Van Horn, J.D., Drews, M.K., Somerville, L.H., Sheridan, M.A., Santillana, R.M., Snyder, J., Hedden, T., Shaw, E.E., Hollinshead, M.O., Renvall, V., Zanzonico, R., Keil, B., Cauley, S., Polimeni, J.R., Tis- dall, D., Buckner, R.L., Wedeen, V.J., Wald, L.L., Toga, A.W., Rosen, B.R., 2016. MGH- USC human connectome project datasets with ultra-high b -value diffusion MRI. Neu- roimage 124, 1108–1114. doi: 10.1016/j.neuroimage.2015.08.075 .
Fischl, B., Salat, D.H., Busa, E., Albert, M., Dieterich, M., Haselgrove, C., van der Kouwe, A., Killiany, R., Kennedy, D., Klaveness, S., Montillo, A., Makris, N., Rosen, B., Dale, A.M., 2002. Whole brain segmentation: automated label- ing of neuroanatomical structures in the human brain. Neuron 33, 341–355.
doi: 10.1016/s0896-6273(02)00569-x .
Fischl, B., van der Kouwe, A., Destrieux, C., Halgren, E., Ségonne, F., Salat, D.H., Busa, E., Seidman, L.J., Goldstein, J., Kennedy, D., Caviness, V., Makris, N., Rosen, B., Dale, A.M., 2004. Automatically parcellating the human cerebral cortex. Cereb. Cor- tex 14, 11–22. doi: 10.1093/cercor/bhg087 .
Fonov, V.S., Evans, A.C., McKinstry, R.C., Almli, C.R., Collins, D.L., 2009. Unbiased non- linear average age-appropriate brain templates from birth to adulthood. Neuroimage 47, S102. doi: 10.1016/S1053-8119(09)70884-5 .
Fyhn, M., Molden, S., Witter, M.P., Moser, E.I., Moser, M.B., 2004. Spatial representation in the entorhinal cortex. Science 305, 1258–1264. doi: 10.1126/science.1099901 . Hafting, T., Fyhn, M., Molden, S., Moser, M.B., Moser, E.I., 2005. Microstructure of a
spatial map in the entorhinal cortex. Nature 436, 801–806. doi: 10.1038/nature03721 .
