MR-based in vivo hippocampal volumetrics: 1. Review of methodologies currently employed

FEATURE REVIEW

MR-based in vivo hippocampal volumetrics: 1. Review ofmethodologies currently employedE Geuze1,2, E Vermetten1,2 and JD Bremner3,4,5

1Department of Military Psychiatry, Central Military Hospital, Utrecht, The Netherlands; 2Department of Psychiatry, RudolfMagnus Institute of Neuroscience, Utrecht, The Netherlands; 3Departments of Psychiatry and Behavioral Sciences andRadiology, Emory University School of Medicine, Atlanta, GA, USA; 4Center for Positron Emission Tomography, Decatur, GA,USA; 5Atlanta VAMC, Decatur, GA, USA

The advance of neuroimaging techniques has resulted in a burgeoning of studies reportingabnormalities in brain structure and function in a number of neuropsychiatric disorders.Measurement of hippocampal volume has developed as a useful tool in the study ofneuropsychiatric disorders. We reviewed the literature and selected all English-language,human subject, data-driven papers on hippocampal volumetry, yielding a database of 423records. From this database, the methodology of all original manual tracing protocols werestudied. These protocols differed in a number of important factors for accurate hippocampalvolume determination including magnetic field strength, the number of slices assessed andthe thickness of slices, hippocampal orientation correction, volumetric correction, softwareused, inter-rater reliability, and anatomical boundaries of the hippocampus. The findings arediscussed in relation to optimizing determination of hippocampal volume.Molecular Psychiatry (2005) 10, 147–159. doi:10.1038/sj.mp.4001580Published online 31 August 2004

Keywords: hippocampus; MRI; volumetry; methodology; neuropsychiatry

The advance of neuroimaging techniques has resultedin a burgeoning of studies reporting abnormalities inbrain structure and function in a number of neurop-sychiatric disorders. One of the brain structureswhich has been a focus of research is the hippocam-pal formation. Magnetic resonance (MR)-based in vivomeasurement of hippocampal volume is an acceptedtechnique, which has been performed in the aged1

and healthy subjects,2 and has revealed a number ofstructural abnormalities in a variety of neurologicaland psychiatric disorders, such as temporal lobeepilepsy,3 Huntington’s disease,4 Turner’s syndrome,5

Cushing’s disease,6 Down’s syndrome,7 Alzheimer’sdisease (AD),8 mild cognitive impairment,9 schizo-phrenia,10 major depression (MD),11 bipolar disor-der,12 post-traumatic stress disorder (PTSD),13

borderline personality disorder,14 chronic alcohol-ism,15 obsessive–compulsive disorder,16 and panicdisorder.17

The MR-derived hippocampal volumetric techni-que has demonstrated good validity and reproduci-bility,18–20 and accuracy of the measurements has been

shown by MRI volumetric measurement of phantomswith a known volume.18,21 However, studies onhippocampal volume in neuropsychiatric disordersare inconclusive and do not always provide consis-tent results. There are differences in laterality (right orleft), direction (increase or decrease), and degree ofthe hippocampal volumetric changes. For example,smaller bilateral hippocampi in patients with schizo-phrenia have been found by a large number ofresearch groups,22–27 but not by others.28–31 Similarly,several groups found smaller bilateral hippocampi inpatients with PTSD,32,33 whereas others were unableto find significantly smaller hippocampi in PTSD.34–36

In MD, significantly smaller bilateral hippocampalvolumes have been reported by some,11,37,38 but not byothers.39,40

Part of the discrepancy among research findingsmay be attributed to the use of different methods forestablishing hippocampal volume. The accuracy andreproducibility of MRI-based in vivo hippocampalvolume measurements depends on three broad fac-tors, namely image acquisition, postacquisition pro-cessing, and volumetric assessment.19 This paperprovides a discussion of the various methods thatstudies of hippocampal volume use. The technicalaspects of image acquisition and postacquisitionprocessing depend on the technical characteristicsand type of scanner available.

It is not the purpose of this review to presentresearchers with another optimal protocol. Rather,

Received 26 February 2004; revised 29 June 2004; accepted 26July 2004

Correspondence: E Geuze, Department of Military Psychiatry,Central Military Hospital and Department of Psychiatry, RudolfMagnus Institute of Neuroscience, Mailbox B.01.2.06, Heidelber-glaan 100, 3584 CX Utrecht, The Netherlands.E-mail: [email protected]

Molecular Psychiatry (2005) 10, 147–159& 2005 Nature Publishing Group All rights reserved 1359-4184/05 $30.00

www.nature.com/mp

this is intended as a review of some of the importantfactors in which these protocols diverge, and then topresent recommendations for optimizing hippocam-pal volume analysis.

Materials and methods

We performed a Medline Indexed search with thekeywords ‘hippocampus,’ ‘volume,’ and ‘MRI. ’ Fromthis database, all English-language, human subject,data-driven papers were selected yielding a databaseof 423 records (only papers published before Decem-ber 31, 2003 were included). Reviews, case studies,and volumetry studies using CT were all excluded.We have assessed the methodology sections of allthese papers, to determine if the paper refers tomethods used by other studies, in order to come upwith the original protocols. This yielded a databaseof approximately 115 ‘original’ protocols. Onlyprotocols in which the manual tracing method(with or without the simultaneous use of regiongrowing or thresholding) was used were includedin this database. Manual tracing protocols constitutethe vast majority of the protocols and are used by 90%of the studies on hippocampal volume in ourdatabase. For this reason, and because both point-counting methods (eg MacFall et al41 and Mackay etal42) and voxel-based morphometric methods (egWright et al31) are different analysis techniques,which are judged by a different set of criteria, theyare difficult to compare to the manual tracingprotocols. Therefore they are not included in thisreview. Nevertheless, although the methodologicaldifferences in these protocols are not mentioned inthis paper, the results from these studies are dis-cussed in the companion paper ‘MR-based in vivohippocampal volumetrics II: Volumetric estimates inneuropsychiatric disorders’.

Results

One of the first general findings that emerges fromthis analysis is that there is a wide range in theamount of reported detail about methodology.Whereas some protocols provide clear data-acquisi-tion and data-processing parameters, as well asdetailed anatomical criteria, a larger number ofpublications do not provide a great amount of detail,making it difficult to compare studies. The protocolsmay differ in a number of factors related to imageacquisition, image processing, and anatomical guide-lines, which are important for accurate hippocampalvolume determination, namely image acquisitionparameters, magnetic field strength, the number ofslices assessed and the thickness of slices, hippo-campal orientation correction, volumetric correction,software used, inter-rater reliability, and anatomicalboundaries of the hippocampus.43–45 These differ-ences are discussed in greater detail below.

Image acquisitionThe protocols employ a wide array of acquisitionsequences. In all, 35% of the protocols use a three-dimensional (3D)-spoiled gradient echo-recalled se-quence (3D SPGR), 15% use a 3D magnetizationprepared rapid acquisition gradient echo (3D-MPRAGE) sequence, 11% use a spin echo (SE)sequence, 7% use an inversion recovery (IR) se-quence, 7% use some other type of gradient-recalledecho (GRE) sequence, 6% use a fast low-angle shot(FLASH) sequence, 4% use some other type of fastfield echo sequence, 4% use a fast SE (FSE) sequence,3% use some other type of echo sequence, 2% usesome other type of acquisition sequence, and 6% donot mention the acquisition sequence used. Inaddition, parameters affecting signal-to-noise ratioand contrast, such as repetition time (TR), echo time(TE), flip angle, field of view, matrix size, and slicethickness vary greatly from study to study. Theprotocols make use of General Electric (52%), Sie-mens/CTI (26%), Philips (12%), Picker (3%), Toshiba(1%), and Ansaldo (1%) scanners. Of the protocols,5% do not mention the manufacturer of their scanner.

Most, (88%), of the protocols used a 1.5T scanner,4% of the protocols mention using a 1T scanner, 3%scanned at 0.5 T, and 3% used a scanner with amagnetic field strength below 0.5 T. Several protocols(2%) used a scanner operating at a magnetic fieldstrength greater than 1.5 T. Bartzokis et al46 havecompared the volumetry of different brain structuresat 0.5 and 1.5 T and demonstrated good interscannerreliability. Although images acquired on the 0.5 Tscanner were acquired using a similar sequence, theydiffered in quality and tissue T2 relaxation times.47

Similarly, although measurement error is lower andmeasurement reliability is improved at 3 T due toincreased tissue contrast, this is not significantlydifferent from that at 1.5 T and does not dramaticallyincrease at 3 T; the increased field strength does notsignificantly affect the volume measurement.48 How-ever, there has also been one report which comparedimages of the hippocampus at 1.5 T to imagesacquired at 4 T.49 Using a slightly different imagingsequence at 1.5 and 4 T, they found that high-resolution imaging provided superior volumetry aswell as an ability to visualize subregions of thehippocampus (JA Detre, personal communication,2003).49 Optimization of image acquisition parametersin combination with increased field strength maythus provide superior contrast and improved hippo-campal volumetry.

Not all of the studies report exactly how manyslices they have assessed, but they do mentionwhether they assessed the whole hippocampus, partof the hippocampus (body or head), or the wholeamygdala–hippocampal complex. In the past, anumber of researchers13,50 have used the body of thehippocampus to evaluate its volume, as this correlateswith total hippocampal size.51 Lower resolution inearly studies also made it difficult to see theamygdala–hippocampal boundary. Currently, however,

MR-based in vivo hippocampal volumetricsE Geuze et al

148

Molecular Psychiatry

measurements of the body of the hippocampus onlyare not acceptable, as this seriously affects the facevalidity of the volumetric measurements. Othersmeasured the tail and body of the hippocampus butdid not include the head;52 Jack et al53 measured thehead and the body of the hippocampus but excludedthe tail. Some researchers, such as Shenton et al,54

measured the amygdala–hippocampal complex,whereas others reliably differentiated between amyg-dala and hippocampus.20,21,55 Of the protocols in thisdatabase, 80% attempted to measure as much of thehippocampus as they could, and included the headand body in their measurements. Only a smallminority of these studies excluded the tail. In all,16% of the protocols measured the whole amygdala–hippocampal complex, and 4% measured the body ofthe hippocampus only.

Image acquisition protocols change rapidly, astechnology advances. At one time state-of-the-artMRI incorporated contiguous 5 mm thick slices;56

however, lately contiguous slices of 1.5 mm or lessare commonly used.57,58 Thus, although Watson et al20

used 3 mm thick slices, later studies performed bythis research group59,60 report using slice thicknessesof 1.5 mm.

The number of slices assessed during hippocampalvolumetry is a variable that is not reported very often,although a good inference of this variable can be madefrom the slice thickness that is used, a variable that isalways reported. The number of slices assessedduring a typical session will vary inversely with thethickness of the slice. Thus, using thicker slicesimplies that fewer slices have been assessed, unlessthe images have been reformatted and resliced usingcomputer software. Using thicker (and thus fewer)slices is less time consuming, and may in some casesbe preferable to using thinner slices.

Image processingThe hippocampi are variably tilted; thus, ideal imagecollection involves perpendicular acquisition of MRimages.56,61 Although such an acquisition is fairlystraightforward with 2D acquisition sequences, 3Dacquisition sequences perpendicular to the hippo-campal axis are impossible to perform on a substan-tial number of MR units.62 Alternatively, this type ofacquisition may also be attained by tilting thepatient’s head, at the expense of increasing patientdiscomfort. It is also possible to reformat the acquiredimages perpendicular to the axis of the hippocampalformation using computer software.

Of the 115 protocols, 39% use various acquisitionsbut reformat the slices at an angle perpendicular tothe long axis of the hippocampal formation. A total of32% do not mention which acquisition orientationthey used, or if they used reformatted images, 22%report acquisitions perpendicular to the AC–PC linewithout reformatting of images, 5% report acquisi-tions perpendicular to the Sylvian fissure, and 3%reported using a head-tilt acquisition. Although thereis no proof that these different acquisition protocols

result in systematic over- or underestimation ofabsolute hippocampal volume,43 these protocolsachieve statistically significant different results.62

There are a number of different software packagesavailable for manual tracing. Almost all of thesoftware that is used employs a combination ofthresholding, manual tracing, and sometimes regiongrowing. The diversity of software packages that isused is so large that it would be too much to dwell onthe differences between them in this paper. However,if we look at those software packages that have beenused in more than three protocols, we see thatAnalyze is by far the most popular software packagethat is used (20.0% of the protocols). Other softwarepackages that are commonly used are MIDAS (6.1%),MEASURE (3.1%), NIH Image (2.6%), BRAINS(2.6%), and DISPLAY (2.6%). In all, 27 protocols(23.5%) report using custom or native scanner soft-ware for analyzing their data. Again, a considerableportion of the protocols (14.8%) do not report whichcomputer program they have used. The other 22.6%of the researchers use various other computer pro-grams both commercial and freely distributed. Insome programs (such as BRAINS, MEASURE, andDisplay), researchers are able to view the brain inthree orthogonal (saggital, coronal, and horizontal)planes simultaneously, thus allowing identification ofanatomical boundaries with greater accuracy. Allsoftware packages employ some method of thresh-olding and/or region growing in combination withmanual tracing.

People with large intracranial volumes tend to havelarger brain structures, such as larger ventricles andlarger hippocampi.63,64 Hippocampal volumes shouldthus be corrected for intersubject variation in headsize. Correcting for head-size or whole brain volumeintroduces two separate sources of error and thusproduces measures with lower reliability.65 However,as Mathalon et al66 showed, head-size correction alsoimproves criterion validity and thus produces highercorrelations with age and diagnostic status thanabsolute values do.

In order to control for these factors, Jack et al67

introduced a region of interest normalization bydividing the region of interest by total intracranialvolume, an approach which they borrowed from theCT literature (see Huckman et al68). The majority(34%) of protocols follow Jack et al’s67 example anduses total intracranial volume to correct for inter-subject variation in head size. Another method whichhas been used quite often (21% of the protocols) is touse division by whole brain volume for normal-ization.1,55,69 Surprisingly, a substantial number ofprotocols (34%) do not use a correction factor at all.Although, in some cases, absolute volumes areneeded (in epilepsy research, or when comparingautomatic and manual volumetrics, for example) andthus controlling for head size is not warranted. Asmall number of studies (4%) uses the correlationalmethod70,71 introduced by Jack et al,53 where thecorrected hippocampal volume (HVn) is derived by


149


taking the original hippocampal volume (HVo) andsubtracting the product of the regression line betweenthe hippocampal volume and intracranial volume,and the difference between the individual intracra-nial volume (TIVi) and the mean intracranial volume(TIVmean). HVn¼HVo�GRAD(TIVi�TIVmean). Othercerebral measures, such as whole brain volume oranother cerebral control area may be substituted forthe intracranial volume in this formula as well.72

The main factor determining accuracy of volu-metric measurements by manual tracing and thresh-olding seems to be the reliability of the within-ratermeasurements.18,73 Apparently, individual reproduc-tion of the hippocampal boundaries is consistent, butreliability between observers is difficult to obtaineven if they are using the same anatomical criter-ia.43,74 One study on intra- and interobserver varia-bility provides an interesting illustration of this, andshowed that one of the observers consistently over-estimated hippocampal volume in comparison to theother observer; thus, intraobserver variability wasfairly consistent with the correlation values of 0.88and 0.97 as opposed to the interobserver correlationvalues, which ranged from 0.62 to 0.73.75

The inter-rater or intrarater reliability that research-ers achieve varies greatly across studies and rangesfrom 0.64 to 0.99. Of the 115 ‘original’ protocols, 60(52%) report ICCs,76 inter-rater or intrarater reliabilityvalues greater than 0.90. In all, 25 protocols (21%)report values of 0.80–0.89 and 6% of the protocolsreport values lower than 0.80. Still the importance ofreporting reliability values has not been taken to heartby all researchers. A substantial portion of theprotocols (21%) do not report any reliability valuesat all.

Anatomical guidelines

The anatomical guidelines that researchers usevary greatly as well. In our database of 423 studies,we have approximately 60 different anatomicalguidelines. It would not be practical to report allthe anatomical guidelines or their variations thatthese protocols use; however, it is interesting to lookat the variations among the most widely usedprotocols. The most widely used protocols weredefined as those protocols that were used in five ormore studies in the original database of 423 records.These protocols have varying anatomical guidelines,which are summarized in Table 1. The protocol ofJack et al,53 which has been revised in 1994,56 and thatof Watson et al protocol20 are the most popular andare reported to have been used in 31 studies each. Theprotocols of Soininen et al55 and Cook et al21 are twoother important and popular ones, which are alsoused frequently. Together, these 14 protocols accountfor 46% of the hippocampal volumetric studiesperformed to date. The anatomical criteria of thesemajor protocols are used in a few other protocols aswell, and are employed in 51% of the researchstudies.

Discussion

Research groups use a variety of different methods,such as manual volumetrics, voxel-based morphome-try, and stereology (or point counting), to assesshippocampal volumes in various neuropsychiatricpopulations. Manual volumetric assessment of thehippocampus has been denoted the ‘gold standard’,but considerable variation exists among researchstudies and there is no standard protocol or metho-dology to which all researchers adhere. The differ-ences in these protocols have been attributed tovarious disparities in acquisition, postacquisitionprocessing, and anatomical guidelines.

Image acquisition protocols should maximize im-age quality and resolution, and should minimize errorsuch as partial volume effects, image quality, headtilt, plane of view, and movement artefacts. Imageacquisitions should also maximize gray matter–whitematter contrast, as this has been shown to affecthippocampal volumetry.85 The contrast betweendifferent brain tissue types is dependent on the imageacquisition sequences used and may thus influencethe hippocampal measurement.

Studies of hippocampal volumetry use a variety ofdifferent image acquisitions. GRE sequences (such as3D SPGR and FLASH86–88) are popular in the field ofhippocampal volumetry, and have been developed toreduce scanning time. Although eliminating the 1801refocusing pulse allows for a significantly shorter TEand TR, GRE sequences are sensitive to susceptibilityeffects and do not compensate for the chemical shiftbetween water and fat.89 Thus, a number of techni-ques such as the frequency selective prepulse and IRare required to increase contrast.88,90 An IR GREvariant, the MPRAGE acquisition sequence, is alsopopular in the field of hippocampal volumetry.Optimized 3D MPRAGE sequences yield higher whitematter to gray matter signal-to-noise ratios than dooptimized 3D FLASH sequences.91 IR sequences92

provide high contrast images of the brain and moreconsistent hippocampal measurements.85 SE se-quences are commonly used in neuroradiology, andprovide excellent anatomic detail at the expense oflonger scan times.90 FSE acquisitions, based on RAREor HASTE sequences, employ more than one SE, andallow faster imaging than the regular SE sequence,without loss of contrast.90 Speed–accuracy tradeoff isan important issue in research as well. Researchersshould ideally employ image acquisition sequences,which provide high signal-to-noise ratios, and max-imum anatomic detail such as FSE sequences.Magnetization-prepared GRE techniques such asMPRAGE and fast-spoiled gradient radiofrequencyat steady state (GRASS)-prepared sequences alsoprovide good signal-to-noise ratio and are preferredto conventional GRE sequences.

Field strengths of 0.5 T or lower place severe limitson resolution. Researchers should use scanners withfield strengths of 1.5 T or more to ensure accuratehippocampal boundary delineation. Volumetry at 4 T


150


Table 1 Anatomical boundaries of the human hippocampus in representative studies

Reference Imageacquisitionsequence/Scanner

Hippocampalmeasurement

Most anteriorslice

Most posterior slice Medial border Lateralborder

Inferiorborder

Additional notes Normative hippocampalvolume (cm3)

Left Right

Bartzokiset al46,77

3D-spoiledGRASSTR/TE/FA25/5/35GE 1.5 T

Wholehippocampus

Level at whichthe alveusdistinguishes theamygdala fromhippocampus

Level where theinferior and superiorcolliculi are jointlyvisualized

Gray matter of thesubiculum isincluded in themeasurement

Gray/whitematterinterface

Gray/whitematterinterface

Subiculum andalveus included inthe measurement

— —

Bigleret al78

FSETR/TE 500/11GE 1.5 T

Wholehippocampus

Anterior aspectof thehippocampus, orthe uncal recessseparatingamygdala fromthe hippocampus

Two of four criteria:presence of superiorcolliculi, presence ofthe medial pulvinarnucleus, visibility ofthe oblong positionof the hippocampusat the crura of thefornices, presence ofa distinct separationof the temporal hornfrom the atria

Anterior choroidalartery, or the pointat which theboundaries of theambient cistern/choriodal fissure aremost readilyidentified

Medial wallof thetemporalhorn

Notmentioned

Hippocampalvolume of controlsfrom Bigler et al(1995)

2.350 2.470

Bogertset al79

3D FLASHTR/TE 40/15Siemens 1 T

Amygdala–hippocampalcomplex

Level at whichamygdalaacquires ovalshape

Level at whichascending fornixsurrounding thepulvinar becomesdistinct

Border between thesubiculum and theparahippocampalgyrus

Notmentioned

Notmentioned

— —

Bremneret al13

3D-poiledGRASSTR/TE/FA25/5/45GE 1.5 T

Body of thehippocampus

First sliceanterior to thesuperiorcolliculus

Proceed 5contiguous 3 mmslices

Mesial edge of thetemporal lobe

Temporalhorn of thelateralventricle

Include thesubicularcomplex andthe uncalcleft

— —

Convitet al80

SE TR/TE630/20Philips 1.5 T

Neck to tailmeasurement

Level of theanterior marginof the lateralgeniculate body

Level at which theposterior pulvinarbecomes visible

CSF of thechoroidal,hippocampal andtransverse tissues

Medial wallof thetemporalhorn

White matterof theparahippo-campal gyrus,subiculumwas included

— —

Cooket al21

GRE 3DTR/TE/FA35/5/35GE 1.5 T

Wholehippocampus

Level at whichthe alveusdistinguishesamygdala fromthe hippocampus

Slice at which thegreatest length of thefornix becomesvisible

Hippocampal anduncal fissures

Notmentioned

Notmentioned

3.229 3.185

Gieddet al81

3D-poiledGRASSTR/TE/FA24/5/4GE 1.5 T

Wholehippocampus

Coronal slicecontaining themost anteriorportions of themammillarybodies

Slice in which thefibers of the fornixare still visible

Not mentioned Notmentioned

Notmentioned

Cornu ammonis,dentate gyrus, andsubiculum includedin the measurement

— —

Honeycuttet al2,82

3D MPRAGETR/TE/FA11/4/15Siemens 1.5 T

Wholehippocampus

Level at whichthe alveusdistinguishesamygdala fromthe hippocampus

Slice where thefornix is visible



White matterof theparahippo-campal gyrus

Alveus included inthe measurement

— —

MR-based

invivo

hippocampalvolum

etricsE

Geuze

etal

151

Mo

lecu

lar

Psy

chia

try

Jacket al53,56

3D SPGRTR/TE/FAmin/min/45GE 1.5 T

Wholehippocampus

Level at whichthe uncal recessof the temporalhorn, or thealveus is visible

Slice where thecrura of the fornicesare seen in fullprofile

CSF in the uncal andambient cistern

CSF in thetemporalhorn

Gray/whitematterjunctionbetween thesubiculumand the whitematter in theparahippo-campal gyrus

Adopted Watsonet al’s (1992)posterior boundarydefinition of thehippocampus in1994

2.400 2.800

Shentonet al54

3D FT-SPGRTR 35GE 1.5 T

Amygdala–hippocampalcomplex

White mattertract linking thetemporal lobewith rest of brain

Slice in which thefibers of the fornixare still visible

Not mentioned Notmentioned

Notmentioned

Mamillary bodiesused to separateamygdala andhippocampus

2.400(anteriorhippo-

campus)

—

Soininenet al55


Wholehippocampus

Level at whichthe head of thehippocampusfirst appearsbelow theamygdala

Slice in which thecrura of the fornicesdepart from thelateral wall of thelateral ventricles/fornices notincluded

Medial wall of thelateral ventricle/subiculum anddentate gyrusincluded

Notmentioned

Uncal portionof the dorsalhippocampusincluded

3.353 3.714

VanPaesschenet al83


Wholehippocampus

Where themamillary bodiesare present/thealveus used as aboundary

First slice where thefornix is visible



White matterof theparahippo-campal gyrus

From the first threeslices one waschosen at randomand from that sliceevery third slice wasmeasuredsystematically

3.320 3.330

Watsonet al20

3D GRETR/TE/FA75/16/60Philips 1.5 T

Wholehippocampus

The CSF in theuncal recess ofthe temporalhorn, whenvisible, is themost reliableboundarybetween thehippocampalhead and theamygdala, if notvisible, thealveus may beused, if neither isvisible, then astraight line isdrawnconnecting theplane of theinferior horn of

Slice where thecrura of the fornicesare seen in fullprofile



Include thesubicularcomplex andthe uncalcleft with theborderseparatingthe subicularcomplex fromtheparahippo-campal gyrus

Subicular complex,dentate gyrus,alveus, and fimbriaincluded inmeasurement. Theseare the most popularanatomical criteriawhich are used by15% of the studies

4.903 5.264

Table 1 (continued)

Reference Imageacquisitionsequence/Scanner

Hippocampalmeasurement

Most anteriorslice

Most posteriorslice

Medial border Lateralborder

Inferiorborder

Additional notes Normative hippocampalvolume (cm3)

Left Right

MR-based

invivo

hippocampalvolum

etricsE

Geuze

etal

152

Mo

lecu

lar

Psy

chia

try

is more sensitive in detecting hippocampal atrophythan at 1.5 T.49 Images of the human hippocampus at7 T even allow researchers to make some distinctionbetween hippocampal layers.93 In the future, increas-ing magnetic field strengths with superior spatial andtemporal resolution and increased signal-to-noiseratio will allow better delineation of the anatomicalboundaries of the hippocampus, with resultantimprovements in accuracy and reliability.

Future research studies on hippocampal volumeshould also make use of thin contiguous slices, sincethey are less likely to be affected by a single falseestimate.57 In 1997, Laakso et al57 examined the effectof slice thickness. They studied 10 normal subjectsand acquired 3D contiguous coronal images with aslice thickness of 1.5–2 mm, which they reformattedinto 1, 3, and 5 mm slices oriented perpendicular tothe hippocampal axis. The hippocampal volumesacquired did not differ significantly between thedifferent slice thicknesses used. Currently, research-ers should no longer make use of 5 or 3 mm slices. AsLaakso et al57 have recommended earlier, thinnerslices should be used, since they are less affected by asingle false estimate.

Visualization of the hippocampus perpendicular toits long axis improves the reliability and reproduci-bility of measurements.18,43,46,57,61,62,77 The majority ofthe studies do not properly report as to whichacquisition orientation they used. Of those who doprovide sufficient detail, the majority report usingimages reformatted perpendicularly to the long axis ofthe hippocampal formation.56 A substantial numberof protocols use different acquisition sequences(either perpendicular to the AC–PC line25,94,84 or theSylvian fissure3,20), but do not reformat their images,and a very small number of studies employ head-tiltprotocols.95–97 3D imaging techniques allow research-ers to save valuable scan time by eliminating the needfor pilot scans needed for consistent positioning ofimages based on internal landmarks, and accomplish-ing this after the scan using multiplanar imagereconstruction capabilities.77

Although Sullivan et al98 were unable to find aneffect of slice orientation, this is probably becausethey only assessed the effect of the APC–hippocam-pus angle (defined as the angle variation of thelongitudinal axis of the hippocampus relative to theAC–PC line) on hippocampal volume. Hasboun et al62

have reliably demonstrated the use of reformattedimages as opposed to nonreformatted images oracquisition by using a head-tilt result in statisticallydifferent hippocampal volumetric estimates; thus, itmust be emphasized that researchers should providesufficient detail in their study design, mentioningwhich method they used. In a study addressingvarious aspects of amygdala and hippocampal volu-metric measurement, Kates et al44 revealed thatrotating images perpendicular to the long axis of thehippocampal formation resulted in a significantlyhigher intrarater reliability in measuring the hippo-campus. In contrast to Hasboun et al,62 Kates et al44

the

late

ral

ven

tric

lew

ith

the

surf

ace

of

the

un

cu

sZ

ipu

rsky

et

al8

4

ME

FC

CG

TR

/TE

2800/

40,8

0G

E1.5

T

Wh

ole

hip

pocam

pu

sS

lice

wh

ere

hip

pocam

pu

sw

as

cle

arl

yd

isti

ngu

ish

ed

from

the

am

ygd

ala

On

esl

ice

(3m

m)

an

teri

or

toth

eim

age

wh

ere

the

vert

ical

fiss

ure

sof

the

Sylv

ian

fiss

ure

are

no

lon

ger

pre

sen

t

Th

ere

gio

nal

ou

tlin

eat

the

ch

oro

idal

fiss

ure

Not

men

tion

ed

Th

ein

terf

ace

of

the

hip

po-

cam

pal

tiss

ue

an

dp

ara

hip

po-

cam

pal

gyru

sw

hit

em

att

er

Th

ism

eth

od

exclu

des

the

most

post

eri

or

regio

nof

the

hip

pocam

pal

bod

yan

dta

il

1.9

90

2.0

70

GR

AS

S¼

gra

die

nt

rad

iofr

equ

en

cy

at

stead

yst

ate

;F

SE¼

fast

spin

ech

o;F

LA

SH¼

fast

low

-an

gle

shots

;S

E¼

spin

ech

o;M

PR

AG

E¼

magn

eti

zati

on

-pre

pare

dra

pid

gra

die

nt

ech

o;

SP

GR¼

spoil

ed

gra

die

nt-

recall

ed

ech

o;

FT

-SP

GR¼

Fou

rier

tran

sform

-sp

oil

ed

gra

die

nt-

recall

ed

ech

o;

ME

FC

CG¼

mu

ltie

ch

ofl

ow

-com

pen

sate

dcard

iac

gate

d;

TR¼

rep

eti

tion

tim

e(m

s);

TE¼

ech

oti

me

(ms)

;F

A¼

flip

an

gle

;m

in¼

min

imal.


153


did not find any significant difference in hippocam-pal volumes obtained with images oriented perpen-dicular to the long axis of the hippocampus, or imagesoriented perpendicular to the AC–PC line. Bartzokiset al77 compared scan–rescan reliability as well asintrarater reliability and found that reformatted 3Dimages showed significantly less scan–rescan varia-bility than nonreformatted images, without sacrificingintrarater reliability. Using reformatted images de-creases the sample size required to detect volumetricchanges by a factor of two. Researchers should thusideally employ images reformatted perpendicular tothe long axis of the hippocampal formation in order toincrease scan–rescan reliability and improve visuali-zation of the hippocampus.

Custom software remains popular in the researchworld. Technological changes occur very rapidly andit is easier to implement these changes if you employcustom software. All software packages utilize somemethod of thresholding and/or region growing incombination with manual tracing. Identification ofanatomical boundaries is more accurate if researchersare able to view the brain in three orthogonal planessimultaneously. Small differences in the algorithmsused to calculate the volume may account for some ofthe variation in the volumes derived in the studies,although there is no reason to assume that thesedifferences are significant. The algorithms that theseprograms use for these functions as well as for volumecalculation/derivation are not similar as well, so thismay also account for some of the differences inresearch findings. However, again, no empiricalevidence exists indicating that this leads to signifi-cant differences.

People with large intracranial volumes tend to havelarger brain structures, such as larger ventricles andlarger hippocampi.63,64 The two major methods tocontrol for intersubject variation in head size aredivision of the region of interest by total intracranialvolume67 or division by whole brain volume.1 Free etal72 investigated several control regions for theirrelationship to hippocampal volume, including thecorpus callosum, the cranial area, parenchymal areaon midsagittal sections, the area of the brain stem onan axial section, and cranial volume and cerebralvolume taken from nine coronal sections throughoutthe cerebrum. The strongest correlation was betweenthe cerebral volume and hippocampal volume. How-ever, correction via the covariance method introducedby Jack et al53 was superior to correction by division,resulted in a greater reduction in variance, andincreased identification of hippocampal sclerosis inpatients with TLE. Correction through division bywhole brain volume is more effective than division bytotal intracranial volume, as the total intracranialvolume remains constant with age, whereas totalbrain volume decreases.99 Several studies have alsoshown that total brain volume is a significantpredictor of subcortical volumes.72,100,101 An impor-tant study by Bigler et al102 revealed that hippocampalvolumes corrected with whole brain volume rather

than total intracranial volume provide greater speci-ficity and sensitivity.

The reliability of measurements and scan–rescanreproducibility of hippocampal volume measurementresearch is a source of major interstudy measurementvariability.43 The reductions found in the variousdisorders are usually small and change little overtime; thus, careful measurements that are reproduci-ble should be made at all times.77 In all studies,regardless of whether one or more raters are used,inter-rater and intrarater reliability values equal to orgreater than 0.9 should be attained for the hippocam-pus. Prospective studies should also employ a similarprotocol at all times even though better criteria existseveral years after the original study was performed.Research has demonstrated that MRI-derived hippo-campal volumes may be reliably acquired in differentresearch centers.103

As becomes evident from Table 1, hippocampalboundaries differ quite a lot among the majorprotocols. There is also considerable variation in theway researchers describe the hippocampal borders.Whereas some provide accurate descriptions sup-ported with pictures and diagrams, others are verymeagre in their account of what they consider to bethe hippocampus. Although, previously, a reliabledistinction between the amygdala and hippocampuswas difficult, due to technical limitations, currentlythere is no empirical reason to warrant not measuringthe structures separately. Measurements of the hippo-campus should include the hippocampus only andshould not be carried out on the hippocampal–amygdala complex as a whole. Studies of fearconditioning have shown that the amygdala plays acritical role in linking external stimuli to defenseresponses, especially those associated with fear.104 Inaddition, the amygdala is a site for some aspects ofemotional memory and modulates memory-relatedprocesses in the hippocampus.105 In many of themanual tracing protocols, the amygdala is measuredin addition to the hippocampus. The variability involumetric studies of the amygdala is less than in thefield of hippocampal volumetry, but striving for moreconsistency in that field should also be encouraged.Illness affects the hippocampus and amygdala differ-ently, and researchers should measure the structuresseparately to obtain a more accurate picture of themorphological changes underlying neuropsychiatricdisorders.

Whereas some protocols include the alveus in theirconception of the hippocampus,20,46,106 others chooseto ignore the alveus.81 Strictly speaking, the alveus isa white matter tract containing axons from hippo-campal, subicular, and septal neurons.107 To avoidconfusion, it may be best to include it in its entirety. Alarge number of the protocols use the alveus toseparate the hippocampus from the amygdala. Othersuse the mamillary bodies to separate amygdaloid andhippocampal tissue.54 Similarly, the protocols differin their inclusion of the subiculum and the uncalcleft, which seriously affects the comparability of


154


these protocols. Inclusion of the subiculum mayincrease the volume of the hippocampus by as muchas 15%. Measuring the tail of the hippocampus is themost difficult part, but using the coronal section onwhich the crux of the fornices is seen in full profile,allow measurement of the head, body, and most of thetail of the hippocampus (90–95%).20,56 Severalauthors define the posterior border of the hippocam-pus as the crura of the fornices,20,21,53,56,108 but othersuse the presence of the inferior and superior colli-culli,46,78 or the absence of the vertical fissures of theSylvian fissure instead.84 All of these differinganatomical boundary definitions are a source of majorvariation among the protocols and constitute thelargest source of discrepancy in normative hippo-campal volumes found by the various researchstudies. Proper referencing to a detailed descriptionof the anatomical criteria used, or complete descrip-tions of the criteria used should be included by allresearchers.

Besides the issues mentioned above, other factorssuch as developmental and gender aspects also affecthippocampal volume. Hippocampal volumetric stu-dies should use proper control groups, which arematched for handedness, IQ, gender, and age. Szaboet al109 showed that right-to-left volume ratios differedsignificantly between right- and left-handed partici-pants for both the amygdala and hippocampus. Full-scale IQ and explicit memory are significantly relatedto hippocampal volume.110–112 Hippocampal volumesare also subject to gender differences. Several studieshave shown113 that the volume of the hippocampalformation is larger in men than in women.1,72,114 Indeveloping children aged 4–18 years, the hippocam-pus increases with age.115 In men, the hippocampusdeclines with age, starting in the third life decade.116

From the age of 54 years hippocampal volume startsto decline at an increased rate (compared to totalbrain atrophy) in both men and women.117 Thesefactors should also be taken into account whencomparing the results found in different studies.

Although manual volumetry is still one of the mostpopular methods to determine hippocampal volumes,automated methods are coming into vogue as well.One of the most troubling aspects of manual tracing isthe subjective interpretation of anatomic variations.As early as 1993, Colombo et al28 introduced anautomated method for determining the volume of theamygdala–hippocampal complex. Voxel-based mor-phometry is an automatic method, which is gainingpopularity and has been used to determine hippo-campal morphometric changes;31,118–127 however,these do not provide absolute hippocampal volumes.

Other automated methods that do provide absolutehippocampal volumes are being developed at a rapidpace. The Knowledge-Guided MRI analysis programis one such program, which uses a combination ofpixel intensity and spatial relationship of atomicstructures to derive hippocampal volume.128 A similarmethod introduced by Ashton et al129 makes use ofgray-scale and edge-detection algorithms as well as

some a priori knowledge in determining hippocampalvolume. The method proposed by Webb et al130

involves warping an atlas (obtained by manualvolumetrics of 30 individuals) to the individual MRimage. Another important automated method is themethod used by Haller and co-workers,131–134 whichuses a high-dimensional fluid transformation to warpa template of the hippocampus and surroundinganatomical structures to an individual MR image.This method has also been validated, and was foundto have less variability than manual tracing.135

Regional fluid registration of serial MRI to investigatebrain change has also been shown to have superiorscan–rescan volumetric consistency; the mean abso-lute volume difference between manual and auto-matic methods was 0.7%.136 However, not all of thesedeformable shape methods take normal hippocampalshape variation into account. Using a deformableshape method, which combines geometric propertiesof hippocampal boundaries, statistical characteriza-tion of normal shape variation, and manually definedboundary points, Shen et al137 demonstrated excellentagreement between automatic and manual volu-metrics of the hippocampus. Shenton et al138 haveused an active, flexible deformable shape model forthe automatic volumetrics of the amygdala–hippo-campal complex to investigate volumetric changes inschizophrenia. These automated methods mark theonset of a new era in structural neuroimaging. It willnot be long before manual volumetrics is replaced byautomatic volumetric methods, which produce simi-lar but more consistent results. This will also make iteasier to implement a common methodology,although the various opinions on automated volu-metric methods that exist today will very likelycontinue their existence far into the future.

Future directions

An appreciation of the differences in researchmethodology helps to understand discrepancies inresearch findings. Ideally, researchers would adopt auniversal methodology. This would lead to moreconsistent results in neuropsychiatric studies ofhippocampal volume, and allow researchers to com-pare results of different studies. However, diversity,fuelled by healthy scepticism is inevitably part of theadvancement of science. Automated volumetrics,which has already found widespread use in variousother brain structures, may also play an importantrole in the field of hippocampal volumetry. Untilthen, manual tracing remains the gold standard.

Acknowledgements

This work was supported by the Dutch Ministry ofDefence, the National Institute of Mental Health R01MH56120, a Veterans Affairs Career DevelopmentAward, and the National Center for Post-traumaticStress Disorder Grant awarded to Dr Bremner.


155


References

1 Bhatia S, Bookheimer SY, Gaillard WD, Theodore WH. Measure-ment of whole temporal lobe and hippocampus for MRvolumetry: normative data. Neurology 1993; 43: 2006–2010.

2 Honeycutt NA, Smith CD. Hippocampal volume measurementsusing magnetic resonance imaging in normal young adults.J Neuroimaging 1995; 5: 95–100.

3 Cendes F, Andermann F, Gloor P, Evans A, Jones-Gotman M,Watson C et al. MRI volumetric measurement of amygdala andhippocampus in temporal lobe epilepsy. Neurology 1993; 43:719–725.

4 Rosas HD, Koroshetz WJ, Chen YI, Skeuse C, Vangel M,Cudkowicz ME et al. Evidence for more widespread cerebralpathology in early HD: An MRI-based morphometric analysis.Neurology 2003; 60: 1615–1620.

5 Murphy DG, DeCarli C, Daly E, Haxby JV, Allen G, White BJ et al.X-chromosome effects on female brain: a magnetic resonanceimaging study of Turner’s syndrome. Lancet 1993; 342:1197–1200.

6 Starkman MN, Gebarski SS, Berent S, Schteingart DE. Hippo-campal formation volume, memory dysfunction, and cortisollevels in patients with Cushing’s syndrome. Biol Psychiatry 1992;32: 756–765.

7 Raz N, Torres IJ, Briggs SD, Spencer WD, Thornton AE, Loken WJet al. Selective neuroanatomic abnormalities in Down’s syndromeand their cognitive correlates: evidence from MRI morphometry.Neurology 1995; 45: 356–366.

8 Chetelat G, Baron JC. Early diagnosis of Alzheimer’s disease:contribution of structural neuroimaging. Neuroimage 2003; 18:525–541.

9 Jack Jr CR, Petersen RC, Xu YC, O’Brien PC, Smith GE, Ivnik RJ etal. Prediction of AD with MRI-based hippocampal volume inmild cognitive impairment. Neurology 1999; 52: 1397–1403.

10 Heckers S. Neuroimaging studies of the hippocampus in schizo-phrenia. Hippocampus 2001; 11: 520–528.

11 Sheline YI, Wang PW, Gado MH, Csernansky JG, Vannier MW.Hippocampal atrophy in recurrent major depression. Proc NatlAcad Sci USA 1996; 93: 3908–3913.

12 Strakowski SM, DelBello MP, Sax KW, Zimmerman ME, ShearPK, Hawkins JM et al. Brain magnetic resonance imaging ofstructural abnormalities in bipolar disorder. Arch Gen Psychiatry1999; 56: 254–260.

13 Bremner JD, Randall P, Scott TM, Bronen RA, Seibyl JP, South-wick SM et al. MRI-based measurement of hippocampal volumein patients with combat-related posttraumatic stress disorder. AmJ Psychiatry 1995; 152: 973–981.

14 Driessen M, Herrmann J, Stahl K, Zwaan M, Meier S, Hill A et al.Magnetic resonance imaging volumes of the hippocampusand the amygdala in women with borderline personalitydisorder and early traumatization. Arch Gen Psychiatry 2000;57: 1115–1122.

15 Sullivan EV, Marsh L, Mathalon DH, Lim KO, Pfefferbaum A.Anterior hippocampal volume deficits in nonamnesic, agingchronic alcoholics. Alcohol Clin Exp Res 1995; 19: 110–122.

16 Kwon JS, Shin YW, Kim CW, Kim YI, Youn T, Han MH et al.Similarity and disparity of obsessive–compulsive disorder andschizophrenia in MR volumetric abnormalities of the hippocam-pus–amygdala complex. J Neurol Neurosurg Psychiatry 2003; 74:962–964.

17 Vythilingam M, Anderson ER, Goddard A, Woods SW, Staib LH,Charney DS et al. Temporal lobe volume in panic disorder—aquantitative magnetic resonance imaging study. Psychiatry Res2000; 99: 75–82.

18 Jack Jr CR, Bentley MD, Twomey CK, Zinsmeister AR. MRimaging-based volume measurements of the hippocampal forma-tion and anterior temporal lobe: validation studies. Radiology1990; 176: 205–209.

19 Pantel J, O’Leary DS, Cretsinger K, Bockholt HJ, Keefe H,Magnotta VA et al. A new method for the in vivo volumetricmeasurement of the human hippocampus with high neuroanato-mical accuracy. Hippocampus 2000; 10: 752–758.

20 Watson C, Andermann F, Gloor P, Jones-Gotman M, Peters T,Evans A et al. Anatomic basis of amygdaloid and hippocampal

volume measurement by magnetic resonance imaging. Neurology1992; 42: 1743–1750.

21 Cook MJ, Fish DR, Shorvon SD, Straughan K, Stevens JM.Hippocampal volumetric and morphometric studies in frontaland temporal lobe epilepsy. Brain 1992; 115(Part 4): 1001–1015.

22 Bogerts B, Lieberman JA, Ashtari M, Bilder RM, Degreef G, LernerG et al. Hippocampus–amygdala volumes and psychopathologyin chronic schizophrenia. Biol Psychiatry 1993; 33: 236–246.

23 Breier A, Buchanan RW, Elkashef A, Munson RC, Kirkpatrick B,Gellad F. Brain morphology and schizophrenia. A magneticresonance imaging study of limbic, prefrontal cortex, and caudatestructures. Arch Gen Psychiatry 1992; 49: 921–926.

24 Gur RE, Turetsky BI, Cowell PE, Finkelman C, Maany V,Grossman RI et al. Temporolimbic volume reductions in schizo-phrenia. Arch Gen Psychiatry 2000; 57: 769–775.

25 Rossi A, Stratta P, Mancini F, Gallucci M, Mattei P, Core L et al.Magnetic resonance imaging findings of amygdala–anteriorhippocampus shrinkage in male patients with schizophrenia.Psychiatry Res 1994; 52: 43–53.

26 Suddath RL, Christison GW, Torrey EF, Casanova MF, WeinbergerDR. Anatomical abnormalities in the brains of monozygotic twinsdiscordant for schizophrenia. N Engl J Med 1990; 322: 789–794.

27 Velakoulis D, Stuart GW, Wood SJ, Smith DJ, Brewer WJ,Desmond P et al. Selective bilateral hippocampal volume lossin chronic schizophrenia. Biol Psychiatry 2001; 50: 531–539.

28 Colombo C, Abbruzzese M, Livian S, Scotti G, Locatelli M,Bonfanti A et al. Memory functions and temporal-limbicmorphology in schizophrenia. Psychiatry Res 1993; 50: 45–56.

29 Deicken RF, Pegues M, Amend D. Reduced hippocampal N-acetylaspartate without volume loss in schizophrenia. SchizophrRes 1999; 37: 217–223.

30 Torres IJ, Flashman LA, O’Leary DS, Swayze II V, Andreasen NC.Lack of an association between delayed memory and hippocam-pal and temporal lobe size in patients with schizophrenia andhealthy controls. Biol Psychiatry 1997; 42: 1087–1096.

31 Wright IC, McGuire PK, Poline JB, Travere JM, Murray RM, FrithCD et al. A voxel-based method for the statistical analysis of grayand white matter density applied to schizophrenia. Neuroimage1995; 2: 244–252.

32 Gurvits TV, Shenton ME, Hokama H, Ohta H, Lasko NB,Gilbertson MW et al. Magnetic resonance imaging study ofhippocampal volume in chronic, combat-related posttraumaticstress disorder. Biol Psychiatry 1996; 40: 1091–1099.

33 Villarreal G, Hamilton DA, Petropoulos H, Driscoll I, RowlandLM, Griego JA et al. Reduced hippocampal volume and totalwhite matter volume in posttraumatic stress disorder. BiolPsychiatry 2002; 52: 119–125.

34 Bonne O, Brandes D, Gilboa A, Gomori JM, Shenton ME,Pitman RK et al. Longitudinal MRI study of hippocampalvolume in trauma survivors with PTSD. Am J Psychiatry 2001;158: 1248–1251.

35 Fennema-Notestine C, Stein MB, Kennedy CM, Archibald SL,Jernigan TL. Brain morphometry in female victims of intimatepartner violence with and without posttraumatic stress disorder.Biol Psychiatry 2002; 52: 1089–1101.

36 Schuff N, Neylan TC, Lenoci MA, Du AT, Weiss DS, Marmar CRet al. Decreased hippocampal N-acetylaspartate in the absence ofatrophy in posttraumatic stress disorder. Biol Psychiatry 2001; 50:952–959.

37 Frodl T, Meisenzahl EM, Zetzsche T, Born C, Groll C, Jager M etal. Hippocampal changes in patients with a first episode of majordepression. Am J Psychiatry 2002; 159: 1112–1118.

38 Sheline YI, Sanghavi M, Mintun MA, Gado MH. Depressionduration but not age predicts hippocampal volume loss inmedically healthy women with recurrent major depression.J Neurosci 1999; 19: 5034–5043.

39 Vakili K, Pillay SS, Lafer B, Fava M, Renshaw PF, Bonello-CintronCM et al. Hippocampal volume in primary unipolar majordepression: a magnetic resonance imaging study. Biol Psychiatry2000; 47: 1087–1090.

40 von Gunten A, Fox NC, Cipolotti L, Ron MA. A volumetric studyof hippocampus and amygdala in depressed patients withsubjective memory problems. J Neuropsychiatry Clin Neurosci2000; 12: 493–498.


156


41 MacFall JR, Byrum CE, Parashos I, Early B, Charles HC, ChittillaV et al. Relative accuracy and reproducibility of regional MRIbrain volumes for point-counting methods. Psychiatry Res 1994;55: 167–177.

42 Mackay CE, Webb JA, Eldridge PR, Chadwick DW, WhitehouseGH, Roberts N. Quantitative magnetic resonance imaging inconsecutive patients evaluated for surgical treatment of temporallobe epilepsy. Magn Reson Imaging 2000; 18: 1187–1199.

43 Jack Jr CR, Theodore WH, Cook M, McCarthy G. MRI-basedhippocampal volumetrics: data acquisition, normal ranges, andoptimal protocol. Magn Reson Imaging 1995; 13: 1057–1064.

44 Kates WR, Abrams MT, Kaufmann WE, Breiter SN, Reiss AL.Reliability and validity of MRI measurement of the amygdala andhippocampus in children with fragile X syndrome. PsychiatryRes 1997; 75: 31–48.

45 Pruessner JC, Li LM, Serles W, Pruessner M, Collins DL, Kabani Net al. Volumetry of hippocampus and amygdala with high-resolution MRI and three-dimensional analysis software: mini-mizing the discrepancies between laboratories. Cereb Cortex2000; 10: 433–442.

46 Bartzokis G, Mintz J, Marx P, Osborn D, Gutkind D, Chiang Fet al. Reliability of in vivo volume measures of hippocampus andother brain structures using MRI. Magn Reson Imaging 1993; 11:993–1006.

47 Bartzokis G, Aravagiri M, Oldendorf WH, Mintz J, Marder SR.Field dependent transverse relaxation rate increase may be aspecific measure of tissue iron stores. Magn Reson Med 1993; 29:459–464.

48 Briellmann RS, Syngeniotis A, Jackson GD. Comparison ofhippocampal volumetry at 1.5 Tesla and at 3 Tesla. Epilepsia2001; 42: 1021–1024.

49 Levy-Reiss I, Gonzalez-Atavales J, King-Stephens D, French JA,Baltuch G, Bagley L et al. High resolution imaging at 4 Tesla forhippocampal volumetry in temporal lobe epilepsy. Proc Intl SocMagn Reson Med 2000; 8: 15.

50 Kaye JA, Swihart T, Howieson D, Dame A, Moore MM, Karnos Tet al. Volume loss of the hippocampus and temporal lobe inhealthy elderly persons destined to develop dementia. Neurology1997; 48: 1297–1304.

51 Kim JH, Tien RD, Felsberg GJ, Osumi AK, Lee N. MR measure-ments of the hippocampus for lateralization of temporal lobeepilepsy: value of measurements of the body vs the wholestructure. Am J Roentgenol 1994; 163: 1453–1457.

52 Ashtari M, Barr WB, Schaul N, Bogerts B. Three-dimensional fastlow-angle shot imaging and computerized volume measurementof the hippocampus in patients with chronic epilepsy of thetemporal lobe. Am J Neuroradiol 1991; 12: 941–947.

53 Jack Jr CR, Twomey CK, Zinsmeister AR, Sharbrough FW,Petersen RC, Cascino GD. Anterior temporal lobes and hippo-campal formations: normative volumetric measurements fromMR images in young adults. Radiology 1989; 172: 549–554.

54 Shenton ME, Kikinis R, Jolesz FA, Pollak SD, LeMay M, Wible CGet al. Abnormalities of the left temporal lobe and thought disorderin schizophrenia. A quantitative magnetic resonance imagingstudy. N Engl J Med 1992; 327: 604–612.

55 Soininen HS, Partanen K, Pitkanen A, Vainio P, Hanninen T,Hallikainen M et al. Volumetric MRI analysis of the amygdala andthe hippocampus in subjects with age-associated memoryimpairment: correlation to visual and verbal memory. Neurology1994; 44: 1660–1668.

56 Jack Jr CR. MRI-based hippocampal volume measurements inepilepsy. Epilepsia 1994; 35(Suppl 6): S21–S29.

57 Laakso MP, Juottonen K, Partanen K, Vainio P, Soininen H. MRIvolumetry of the hippocampus: the effect of slice thickness onvolume formation. Magn Reson Imaging 1997; 15: 263–265.

58 MacQueen GM, Campbell S, McEwen BS, Macdonald K, AmanoS, Joffe RT et al. Course of illness, hippocampal function, andhippocampal volume in major depression. Proc Natl Acad SciUSA 2003; 100: 1387–1392.

59 Fuerst D, Shah J, Shah A, Watson C. Hippocampal sclerosis is aprogressive disorder: a longitudinal volumetric MRI study. AnnNeurol 2003; 53: 413–416.

60 Watson C, Cendes F, Fuerst D, Dubeau F, Williamson B, Evans Aet al. Specificity of volumetric magnetic resonance imaging

in detecting hippocampal sclerosis. Arch Neurol 1997; 54:67–73.

61 Press GA, Amaral DG, Squire LR. Hippocampal abnormalities inamnesic patients revealed by high-resolution magnetic resonanceimaging. Nature 1989; 341: 54–57.

62 Hasboun D, Chantome M, Zouaoui A, Sahel M, Deladoeuille M,Sourour N et al. MR determination of hippocampal volume:comparison of three methods. Am J Neuroradiol 1996; 17:1091–1098.

63 Synek V, Reuben JR. The ventricular-brain ratio using planimetricmeasurement of EMI scans. Br J Radiol 1976; 49: 233–237.

64 Zatz LM, Jernigan TL. The ventricular-brain ratio on computedtomography scans: validity and proper use. Psychiatry Res 1983;8: 207–214.

65 Arndt S, Cohen G, Alliger RJ, Swayze II VW, Andreasen NC.Problems with ratio and proportion measures of imaged cerebralstructures. Psychiatry Res 1991; 40: 79–89.

66 Mathalon DH, Sullivan EV, Rawles JM, Pfefferbaum A. Correctionfor head size in brain-imaging measurements. Psychiatry Res1993; 50: 121–139.

67 Jack Jr CR, Petersen RC, O’Brien PC, Tangalos EG. MR-basedhippocampal volumetry in the diagnosis of Alzheimer’s disease.Neurology 1992; 42: 183–188.

68 Huckman MS, Fox J, Topel J. The validity of criteria for theevaluation of cerebral atrophy by computed tomography. Radi-ology 1975; 116: 85–92.

69 Pearlson GD, Harris GJ, Powers RE, Barta PE, Camargo EE, ChaseGA et al. Quantitative changes in mesial temporal volume,regional cerebral blood flow, and cognition in Alzheimer’sdisease. Arch Gen Psychiatry 1992; 49: 402–408.

70 Colchester A, Kingsley D, Lasserson D, Kendall B, Bello F,Rush C et al. Structural MRI volumetric analysis in patientswith organic amnesia, 1: methods and comparative findingsacross diagnostic groups. J Neurol Neurosurg Psychiatry 2001; 71:13–22.

71 Kopelman MD, Lasserson D, Kingsley D, Bello F, Rush C,Stanhope N et al. Structural MRI volumetric analysis in patientswith organic amnesia, 2: correlations with anterograde memoryand executive tests in 40 patients. J Neurol Neurosurg Psychiatry2001; 71: 23–28.

72 Free SL, Bergin PS, Fish DR, Cook MJ, Shorvon SD, Stevens JM.Methods for normalization of hippocampal volumes measuredwith MR. Am J Neuroradiol 1995; 16: 637–643.

73 Cascino GD, Jack Jr CR, Parisi JE, Marsh WR, Kelly PJ, SharbroughFW et al. MRI in the presurgical evaluation of patients withfrontal lobe epilepsy and children with temporal lobe epilepsy:pathologic correlation and prognostic importance. Epilepsy Res1992; 11: 51–59.

74 Cendes F, Leproux F, Melanson D, Ethier R, Evans A, Peters T etal. MRI of amygdala and hippocampus in temporal lobe epilepsy.J Comput Assist Tomogr 1993; 17: 206–210.

75 Achten E, Deblaere K, De Wagter C, Van Damme F, Boon P, DeReuck J et al. Intra- and interobserver variability of MRI-basedvolume measurements of the hippocampus and amygdalausing the manual ray-tracing method. Neuroradiology 1998; 40:558–566.

76 Shrout PE, Fleiss JL. Psychological Bulletin 86: 420–428, 479.77 Bartzokis G, Altshuler LL, Greider T, Curran J, Keen B, Dixon WJ.

Reliability of medial temporal lobe volume measurements usingreformatted 3D images. Psychiatry Res 1998; 82: 11–24.

78 Bigler ED, Blatter DD, Anderson CV, Johnson SC, Gale SD,Hopkins RO et al. Hippocampal volume in normal aging andtraumatic brain injury. Am J Neuroradiol 1997; 18: 11–23.

79 Bogerts B, Ashtari M, Degreef G, Alvir JM, Bilder RM, LiebermanJA. Reduced temporal limbic structure volumes on magneticresonance images in first episode schizophrenia. Psychiatry Res1990; 35: 1–13.

80 Convit A, de Leon MJ, Tarshish C, De Santi S, Kluger A, RusinekH et al. Hippocampal volume losses in minimally impairedelderly. Lancet 1995; 345: 266.

81 Giedd JN, Vaituzis AC, Hamburger SD, Lange N, Rajapakse JC,Kaysen D et al. Quantitative MRI of the temporal lobe, amygdala,and hippocampus in normal human development: ages 4-18years. J Comp Neurol 1996; 366: 223–230.


157


82 Honeycutt NA, Smith PD, Aylward E, Li Q, Chan M, Barta PE etal. Mesial temporal lobe measurements on magnetic resonanceimaging scans. Psychiatry Res 1998; 83: 85–94.

83 Van Paesschen W, Connelly A, King MD, Jackson GD,Duncan JS. The spectrum of hippocampal sclerosis: a quantita-tive magnetic resonance imaging study. Ann Neurol 1997; 41:41–51.

84 Zipursky RB, Marsh L, Lim KO, DeMent S, Shear PK, Sullivan EVet al. Volumetric MRI assessment of temporal lobe structures inschizophrenia. Biol Psychiatry 1994; 35: 501–516.

85 Wieshmann UC, Free SL, Stevens JM, Shorvon SD. Image contrastand hippocampal volumetric measurements. Magn Reson Ima-ging 1998; 16: 13–17.

86 Haase A. Snapshot FLASH MRI. Applications to T1, T2, andchemical-shift imaging. Magn Reson Med 1990; 13: 77–89.

87 Haase A, Frahm J, Matthaei W, Haenicke W, Merboldt KD. FLASHimaging, rapid NMR imaging using low flip angle pulses. J MagnReson 67: 256–266, 286.

88 Reiser M, Faber SC. Recent and future advances in high-speedimaging. Eur Radiol 1997; 7(Suppl 5): 166–173.

89 Mugler III JP. Overview of MR imaging pulse sequences. MagnReson Imaging Clin N Am 1999; 7: 661–697.

90 Jackson EF, Ginsberg LE, Schomer DF, Leeds NE. A review of MRIpulse sequences and techniques in neuroimaging. Surg Neurol1997; 47: 185–199.

91 Deichmann R, Good CD, Josephs O, Ashburner J, Turner R.Optimization of 3-D MP-RAGE sequences for structural brainimaging. Neuroimage 2000; 12: 112–127.

92 Bydder GM, Hajnal JV, Young IR. MRI: use of the inversionrecovery pulse sequence. Clin Radiol 1998; 53: 159–176.

93 Wieshmann UC, Symms MR, Mottershead JP, MacManus DG,Barker GJ, Tofts PS et al. Hippocampal layers on high resolutionmagnetic resonance images: real or imaginary? J Anat 1999;195(Part 1): 131–135.

94 Jacobsen LK, Giedd JN, Vaituzis AC, Hamburger SD, RajapakseJC, Frazier JA et al. Temporal lobe morphology in childhood-onset schizophrenia. Am J Psychiatry 1996; 153: 355–361.

95 Buchanan RW, Breier A, Kirkpatrick B, Elkashef A, Munson RC,Gellad F et al. Structural abnormalities in deficit and nondeficitschizophrenia. Am J Psychiatry 1993; 150: 59–65.

96 Coffey CE, Wilkinson WE, Weiner RD, Parashos IA, Djang WT,Webb MC et al. Quantitative cerebral anatomy in depression. Acontrolled magnetic resonance imaging study. Arch Gen Psy-chiatry 1993; 50: 7–16.

97 DeLisi LE, Tew W, Xie S, Hoff AL, Sakuma M, Kushner M et al. Aprospective follow-up study of brain morphology and cognitionin first-episode schizophrenic patients: preliminary findings.Biol Psychiatry 1995; 38: 349–360.

98 Sullivan EV, Marsh L, Mathalon DH, Lim KO, Pfefferbaum A.Age-related decline in MRI volumes of temporal lobe gray matterbut not hippocampus. Neurobiol Aging 1995; 16: 591–606.

99 Mori E, Hirono N, Yamashita H, Imamura T, Ikejiri Y, Ikeda M etal. Premorbid brain size as a determinant of reserve capacityagainst intellectual decline in Alzheimer’s disease. Am JPsychiatry 1997; 154: 18–24.

100 Finlay BL, Darlington RB. Linked regularities in the develop-ment and evolution of mammalian brains. Science 1995; 268:1578–1584.

101 Mueller EA, Moore MM, Kerr DC, Sexton G, Camicioli RM,Howieson DB et al. Brain volume preserved in healthy elderlythrough the eleventh decade. Neurology 1998; 51: 1555–1562.

102 Bigler ED, Tate DF. Brain volume, intracranial volume, anddementia. Invest Radiol 2001; 36: 539–546.

103 Jack Jr CR, Slomkowski M, Gracon S, Hoover TM, Felmlee JP,Stewart K et al. MRI as a biomarker of disease progressionin a therapeutic trial of milameline for AD. Neurology 2003; 60:253–260.

104 LeDoux J. The emotional brain, fear, and the amygdala. Cell MolNeurobiol 2003; 23: 727–738.

105 Richter-Levin G. The amygdala, the hippocampus, and emotionalmodulation of memory. Neuroscientist 2004; 10: 31–39.

106 Spencer SS, McCarthy G, Spencer DD. Diagnosis of medialtemporal lobe seizure onset: relative specificity and sensitivity ofquantitative MRI. Neurology 1993; 43: 2117–2124.

107 Duvernoy HM. The Human Hippocampus: Functional Anatomy,Vascularization, and Serial Sections with MRI, Vol. 98.

108 Steciuk M, Kram M, Kramer GL, Petty F. Acute stress does notalter 5-HT1A receptor density. Prog Neuropsychopharmacol BiolPsychiatry 2000; 24: 155–161.

109 Szabo CA, Xiong J, Lancaster JL, Rainey L, Fox P. Amygdalar andhippocampal volumetry in control participants: differencesregarding handedness. Am J Neuroradiol 2001; 22: 1342–1345.

110 Andreasen NC, Flaum M, Swayze II V, O’Leary DS, Alliger R,Cohen G et al. Intelligence and brain structure in normalindividuals. Am J Psychiatry 1993; 150: 130–134.

111 Foster JK, Meikle A, Goodson G, Mayes AR, Howard M, SunramSI et al. The hippocampus and delayed recall: bigger is notnecessarily better? Memory 1999; 7: 715–732.

112 Chantome M, Perruchet P, Hasboun D, Dormont D, Sahel M,Sourour N et al. Is there a negative correlation between explicitmemory and hippocampal volume? Neuroimage 1999; 10: 589–595.

113 Salmenpera T, Kalviainen R, Partanen K, Mervaala E, Pitkanen A.MRI volumetry of the hippocampus, amygdala, entorhinal cortex,and perirhinal cortex after status epilepticus. Epilepsy Res 2000;40: 155–170.

114 Raz N, Gunning FM, Head D, Dupuis JH, McQuain J, Briggs SD etal. Selective aging of the human cerebral cortex observed in vivo:differential vulnerability of the prefrontal gray matter. CerebCortex 1997; 7: 268–282.

115 Giedd JN, Castellanos FX, Rajapakse JC, Vaituzis AC, RapoportJL. Sexual dimorphism of the developing human brain. ProgNeuropsychopharmacol Biol Psychiatry 1997; 21: 1185–1201.

116 Pruessner JC, Collins DL, Pruessner M, Evans AC. Age andgender predict volume decline in the anterior and posteriorhippocampus in early adulthood. J Neurosci 2001; 21: 194–200.

117 Liu RS, Lemieux L, Bell GS, Sisodiya SM, Shorvon SD, SanderJW et al. A longitudinal study of brain morphometrics usingquantitative magnetic resonance imaging and difference imageanalysis. Neuroimage 2003; 20: 22–33.

118 Burton EJ, Karas G, Paling SM, Barber R, Williams ED,Ballard CG et al. Patterns of cerebral atrophy in dementia withLewy bodies using voxel-based morphometry. Neuroimage 2002;17: 618–630.

119 Job DE, Whalley HC, McConnell S, Glabus M, Johnstone EC,Lawrie SM. Structural gray matter differences between first-episode schizophrenics and normal controls using voxel-basedmorphometry. Neuroimage 2002; 17: 880–889.

120 Karas GB, Burton EJ, Rombouts SA, van Schijndel RA, O’Brien JT,Scheltens P et al. A comprehensive study of gray matter loss inpatients with Alzheimer’s disease using optimized voxel-basedmorphometry. Neuroimage 2003; 18: 895–907.

121 Keller SS, Wieshmann UC, Mackay CE, Denby CE, Webb J,Roberts N. Voxel based morphometry of grey matter abnormal-ities in patients with medically intractable temporal lobeepilepsy: effects of side of seizure onset and epilepsy duration.J Neurol Neurosurg Psychiatry 2002; 73: 648–655.

122 Kubicki M, Shenton ME, Salisbury DF, Hirayasu Y, Kasai K,Kikinis R et al. Voxel-based morphometric analysis of gray matterin first episode schizophrenia. Neuroimage 2002; 17: 1711–1719.

123 Maguire EA, Spiers HJ, Good CD, Hartley T, Frackowiak RS,Burgess N. Navigation expertise and the human hippocampus:a structural brain imaging analysis. Hippocampus 2003; 13:250–259.

124 Mummery CJ, Patterson K, Price CJ, Ashburner J, Frackowiak RS,Hodges JR. A voxel-based morphometry study of semanticdementia: relationship between temporal lobe atrophy andsemantic memory. Ann Neurol 2000; 47: 36–45.

125 Shah PJ, Ebmeier KP, Glabus MF, Goodwin GM. Cortical greymatter reductions associated with treatment-resistant chronicunipolar depression. Controlled magnetic resonance imagingstudy. Br J Psychiatry 1998; 172: 527–532.

126 Woermann FG, Free SL, Koepp MJ, Ashburner J, Duncan JS.Voxel-by-voxel comparison of automatically segmented cerebralgray matter—a rater-independent comparison of structural MRIin patients with epilepsy. Neuroimage 1999; 10: 373–384.

127 Yamasue H, Kasai K, Iwanami A, Ohtani T, Yamada H, Abe Oet al. Voxel-based analysis of MRI reveals anterior cingulate


158


gray-matter volume reduction in posttraumatic stress disorderdue to terrorism. Proc Natl Acad Sci USA 2003; 100: 9039–9043.

128 Gosche KM, Mortimer JA, Smith CD, Markesbery WR, SnowdonDA. An automated technique for measuring hippocampalvolumes from MR imaging studies. Am J Neuroradiol 2001; 22:1686–1689.

129 Ashton EA, Berg MJ, Parker KJ, Weisberg J, Chen CW,Ketonen L. Segmentation and feature extraction techniques, withapplications to MRI head studies. Magn Reson Med 1995; 33:670–677.

130 Webb J, Guimond A, Eldridge P, Chadwick D, Meunier J, ThirionJP et al. Automatic detection of hippocampal atrophy onmagnetic resonance images. Magn Reson Imaging 1999; 17:1149–1161.

131 Cardenas VA, Du AT, Hardin D, Ezekiel F, Weber P, Jagust WJet al. Comparison of methods for measuring longitudinal brainchange in cognitive impairment and dementia. Neurobiol Aging2003; 24: 537–544.

132 Csernansky JG, Joshi S, Wang L, Haller JW, Gado M, Miller JPet al. Hippocampal morphometry in schizophrenia by highdimensional brain mapping. Proc Natl Acad Sci USA 1998; 95:11406–11411.

133 Haller JW, Banerjee A, Christensen GE, Gado M, Joshi S, MillerMI et al. Three-dimensional hippocampal MR morphometry withhigh-dimensional transformation of a neuroanatomic atlas.Radiology 1997; 202: 504–510.

134 Wang L, Joshi SC, Miller MI, Csernansky JG. Statistical analysisof hippocampal asymmetry in schizophrenia. Neuroimage 2001;14: 531–545.

135 Hogan RE, Mark KE, Wang L, Joshi S, Miller MI, Bucholz RD.Mesial temporal sclerosis and temporal lobe epilepsy: MRimaging deformation-based segmentation of the hippocampusin five patients. Radiology 2000; 216: 291–297.

136 Crum WR, Scahill RI, Fox NC. Automated hippocampalsegmentation by regional fluid registration of serial MRI:validation and application in Alzheimer’s disease. Neuroimage2001; 13: 847–855.

137 Shen D, Moffat S, Resnick SM, Davatzikos C. Measuring size andshape of the hippocampus in MR images using a deformableshape model. Neuroimage 2002; 15: 422–434.

138 Shenton ME, Gerig G, McCarley RW, Szekely G, Kikinis R.Amygdala–hippocampal shape differences in schizophrenia: theapplication of 3D shape models to volumetric MR data.Psychiatry Res 2002; 115: 15–35.


159


MR-based in vivo hippocampal volumetrics: 1. Review of methodologies currently employed

Documents

Transcript of MR-based in vivo hippocampal volumetrics: 1. Review of methodologies currently employed