Alzheimer’s & Dementia 7 (2011) 171e174

Perspectives

Harmonization of magnetic resonance-based manual hippocampalsegmentation: A mandatory step for wide clinical use

Giovanni B. Frisonia,*, Clifford R. Jackb

aLENITEM Laboratory of Epidemiology, Neuroimaging and Telemedicine, IRCCS San Giovanni di Dio-Fatebenefratelli, Brescia, ItalybDepartment of Diagnostic Radiology, Mayo Clinic, Rochester, MN, USA

Abstract Hippocampal atrophy is a marker of disease state and progression in Alzheimer’s disease. The gold

*Corresponding au

13.

E-mail address: gf

1552-5260/$ - see fro

doi:10.1016/j.jalz.201

standard to measure hippocampal volume is through manual segmentation. A number of protocols tomeasure hippocampal volume through manual segmentation have been developed, but the markedheterogeneity of anatomical landmarks has given rise to wide variability of volume estimates.With the aim of fostering the use of hippocampal volume in routine clinical settings, an internationaltask force is currently working on developing a harmonized protocol that will resolve and reduce thepresent heterogeneity. The task force will then validate the harmonized protocol, develop harmonizedprobabilistic hippocampal maps, and develop illustrative and educational material on the use of theharmonized protocol and maps.� 2011 The Alzheimer’s Association. All rights reserved.

Keywords: Alzheimer’s disease; Hippocampal volumetry; MR imaging; Early diagnosis; Clinical trials; Outcome measures

1. Introduction

In the fields of Alzheimer’s disease (AD) care and drugdevelopment, there is an urgent need for the developmentof procedures that would help to estimate hippocampal atro-phy accurately and consistently. In the revised criteria for thediagnosis of AD, an estimate of hippocampal atrophy fromstructural magnetic resonance imaging (MRI) is a key sup-portive marker [1]. In patients with AD, hippocampal atro-phy has been measured in clinical trials of tramiprosate,atorvastatin, AN1792, xaliproden, and donepezil (Table 1).Some of these found evidence of a beneficial drug effecton reduction of hippocampal volume, despite variable clini-cal effects. A notable exception is the AN1792 trial in whichtreated subjects lost more hippocampal volume than non-treated patients, although the loss was not significant. Algo-rithms that automatically segment (delineate) thehippocampus from the surrounding brain tissue on MR brainscans are being actively developed; these will require thegold standard of manual segmentation for validation [2,3].The potential future availability of drugs that alter

thor. Tel.: 39-03-03-50-13-61; Fax: 39-03-03-50-13-

risoni@fatebenefratelli.it

nt matter � 2011 The Alzheimer’s Association. All rights r

0.06.007

progression of cognitive deterioration will make a securediagnosis at the earliest possible stages imperative.

Evidence from imaging-pathological correlations showsthat manual hippocampal segmentation is a valid markerof neurodegenerative changes in AD [4], and several studiesin clinical AD populations from laboratories worldwidehave reported that hippocampal volume in patients were15% to 40% smaller (more atrophied) than controls [5].However, different laboratories use different anatomicallandmarks and measurement procedures. This inconsistencyof approach means that estimates of “normal” hippocampalvolumes may differ by a maximum of 2.5-fold (Table 2) [6].A meta-analysis of rates of atrophy over time showed aneven wider range, with rates varying from 0.32% to 6.8%per year [7]. Without comparability of methods, it is impos-sible to determine whether these differences reflect a neuro-biological heterogeneity or how much of this variance isdetermined by the different protocols.

The most validated procedure to estimate hippocampalatrophy is to calculate hippocampal volumes with manualoutlining using anatomical landmarks by an expert rater onhigh resolution T1-weighted MRI [5]. This manual volume-try is also used as the standard against which automatedsegmentation algorithms [8e12] are assessed; however, inthe absence of an agreed reference protocol for manual

eserved.

Table 1

Clinical trials with drugs for Alzheimer’s disease where hippocampal volume measures were included in the study design

Drug Trial name Patients Effect on hippocampal volume loss Segmentation method Reference

AN1792 e AD Change from baseline to follow-up:

Placebo: 22.86% (SD: 3.19)

Treated: 23.78% (SD: 2.63), P 5 .12

Manual [14] 15

Atorvastatin ADCLT AD Change from baseline to follow-up:

Placebo: 2134 (SD: 174) mm3

Treated: 583 (SD: 354) mm3, P . .05

Manual [16,17] 18

LEADe AD Significant smaller annualized decrease in

hippocampal volume in treated patients

Semi-automated MIDAS [19] 20,21

Donepezil Memory

impairment study

MCI Change from baseline to follow-up in e4

carriers:

Placebo: 26.14% (SE: 3.49)

Treated: 24.50% (SE: 2.28), P 5 .07

Manual [22,23] 24

Tramiprosate Alphase AD Change from baseline to follow-up:

Placebo: 2419 (SE: 113) mm3

Treated 100 mg: 2135 (SE: 58) mm3,

P 5 .035

Treated: 150 mg: 79.5 (SE: 133) mm3,

P 5 .009

Not mentioned 25

Xaliproden EFC2724 AD Significantly less hippocampal atrophy in

treated patients

Not mentioned 26

NOTE. Some manuscripts fail to provide details on the magnitude and significance of the effect.

Abbreviations: AD, Alzheimer’s disease; SD, standard deviation; ADCLT, Alzheimer’s disease cholesterol-lowering treatment; LEADe, Lipitor’s effect in

Alzheimer’s dementia; MCI, mild cognitive impairment; SE, standard error.

G.B. Frisoni and C.R. Jack / Alzheimer’s & Dementia 7 (2011) 171e174172

volumetry, the comparison of the accuracy of differentautomated methods is virtually impossible.

An international task force has recently been gathered bythe authors of this article with the aim of developing a harmo-nized protocol that will overcome the present heterogeneity(Table 3). The project will run in four phases. Phase I willharmonize existing protocols, phase II will validate theharmonized protocol, phase IIIwill develop harmonized prob-abilistic hippocampal maps, and phase IV will develop anillustrative and educational material on the use of the harmo-

Table 2

Extreme values of normal hippocampal volumes according to studies using differ

Reference

Most anterior

slice

Most posterior

slice

Medial

border

L

b

14 CSF in uncal

recess of

temporal

horn or

alveus

Slice where the

crura of

fornices are

seen in full

profile

Mesial edge

of the

temporal

lobe

T

27 Slice where

hippocampus is

clearly

distinguished

from the

amygdala

One slice anterior

to where the

vertical Sylvian

fissures are no

longer present

Regional outline at

the level of the

choroidal

fissure

N

Adapted from Geuze et al., 2005 [6].

nized protocol and maps. Under the auspices of the Alz-heimer’s Association, the task force had met first in Chicagoin July 2008 to discuss the study design, and recently in Tor-onto in April 2010, wherework to datewas presented and dis-cussed in a hybrid in-person and remote (webinar) workshop.

Twelve most frequently used protocols for manualhippocampal segmentation were selected from the Alz-heimer’s literature; anatomical landmarkswere extracted; hip-pocampi from two sample brain scans (one representativeAD patient and one healthy control from the Alzheimer’s

ent protocols for manual segmentation

ateral

order

Inferior

border

Hippocampal volume (cm3)

Left Right

emporal horn

of the lateral

ventricle

Includes

subicular

complex and

uncal cleft with

the border

separating the

subicular

complex from

the parahippo

campal gyrus

4.903 5.264

ot mentioned Interface of the

hippocampal

tissue and

parahip-

pocampal

gyrus white

matter

1.990 2.070

Table 3

The expert working group

EADC centres ADNI centres Other centres

Population-based

studies

Statistical working

group Advisors

N Fox, London,

United Kingdom

M Albert, Johns

Hopkins

University,

Baltimore, MD

J Pruessner*,

McGill

University,

QC, Canada

Rotterdam Scan

Study, M B

reteler/T den

Heijer

P Pasqualetti,

AFAR, Roma

EADC P.I.s: B Winbald

and L Froelich ADNI

P.I.:MWeiner,UCSF,CA

Clinical issues:

PJ Visser, Maastricht,

The Netherlands

A Simmons, London,

United Kingdom

J Csernansky*,

Northwestern

University, IL

R Camicioli/N

Malykhin*

University

Alberta, AB,

Canada

PATH through life,

P Sachdev/JJ

Maller

S Duchesne,

Laval

University,

Canada

L-O Wahlund,

Stockholm,

Sweden

M De Leon*,

New York, NY

C Watson*, WSU,

Detroit, MI

L Collins, MNI, McGill,

Montreal

Dissemination

and Education:

G Waldemar,

Copenhagen,

Denmark

F Barkhof/P

Scheltens,

Amsterdam, The

Netherlands

R Killiany*,

Boston USM,

MA

J O’Brien,

Newcastle,

United

Kingdom

GB Frisoni,

Brescia, Italy

G Bartzokis*,

UCLA, CA

Population studies:

L Launer, NIA,

Bethesda, MD

W Jagust,

Berkeley, CA

H Soininen*,

Kuopio, Finland

C DeCarli, UC

Davis, CA

B Dubois/S

Leherici* Paris,

France

CR Jack*, Rochester,

MN

H Hampel/J Pantel*,

University of

Frankfurt, DE

PM Thompson,

LONI, UCLA,

CA

S Teipel, Rostock,

DE

L deToledo-Morrell*,

Rush UMC,

Chicago, IL

J Kaye, Portland,

OR

M Weiner/S

Mueller, UCSF, CA

D Bennett, Rush

ADC, Chicago, IL

*The authors of segmentation protocols that will contribute to the harmonized protocol.

G.B. Frisoni and C.R. Jack / Alzheimer’s & Dementia 7 (2011) 171e174 173

Disease Neuroimaging Initiative [ADNI] database) weresegmented following all of the 12 protocols and the accu-racy of the interpretation of the protocols was checked dur-ing interactive webinars with the protocols’ authors. Theanatomical landmarks certified by the protocols’ authorswere semantically harmonized; the differences were opera-tionalized into tracing units summarizing all the variabilityamong protocols; and 3D visual representations of the trac-ing units were developed. The work done so far has beenpresented at the American Academy of Neurology meeting[13] and the pertinent material can be found at http://www.hippocampal-protocol.net.

This is a preparatorywork for empirical testing of the con-tribution of each tracing unit to segmentation accuracy and

volumetric differences between patients and controls. The re-sults will be provided to a Delphi panel that will reach con-sensus on a harmonized protocol. The harmonized protocolwill be validated with neuropathological data and its accu-racy will be compared with currently used protocols. Finally,hippocampal probability maps will be developed. These willbe instrumental to the development of standard operationalprocedures for the measurement of hippocampal volume,an essential feature of anymedical test to be used in the clinic.

Social awareness and scientific knowledge of AD haveincreased dramatically in the past 20 years. However, thera-peutic options are currently limited to symptomatic drugsand diagnosis is still largely based on individual physicianexperience and subjective judgment. Standard operational

G.B. Frisoni and C.R. Jack / Alzheimer’s & Dementia 7 (2011) 171e174174

procedures for the assessment of disease markers will be thekey to drug discovery and to the development of more effec-tive, technology-assisted care of patients with AD.

Acknowledgments

Marina Boccardi led the technical group of Rossana Gan-zola, Simon Duchesne, Nicolas Robitaille, Alberto Redolfi,Michela Pievani, and Anna Caroli. Enrica Cavedo helped inthe editorial process of this manuscript. This project wasfunded partly from unrestricted grants from Lilly andWyeth.The Alzheimer’s Association has generously taken charge oforganization of workshops. The authors of the protocolshave been key to the work done so far: George Bartzokis,John G. Csernansky, Mony De Leon, Leyla deToledo-Morrell, Ron Killiany, Stephane Lehericy, Nikolai Maly-khin, Johannes Pantel, Jens Pruessner, Hilkka Soininen,and Craig Watson.

References

[1] Dubois B, Feldman HH, Jacova C, Dekosky ST, Barberger-Gateau P,

Cummings J, et al. Research criteria for the diagnosis of Alzheimer’s

disease: revising the NINCDS-ADRDA criteria. Lancet Neurol 2007;

6:734e46.

[2] Morey RA, Petty CM, Xu Y, Hayes JP, Wagner HR II, Lewis DV, et al.

A comparison of automated segmentation and manual tracing for

quantifying hippocampal and amygdala volumes. Neuroimage 2009;

45:855e66.

[3] Chupin M, Hammers A, Liu RSN, Colliot O, Burdett J, Bardinet E,

et al. Automatic segmentation of the hippocampus and the amygdala

driven by hybrid constraints: method and validation. Neuroimage

2009;46:749e61.[4] Bobinski M, de Leon MJ, Wegiel J, Desanti S, Convit A, Saint

Louis LA, et al. The histological validation of post mortem magnetic

resonance imaging-determined hippocampal volume in Alzheimer’s

disease. Neuroscience 2000;95:721e5.[5] Bosscher L, Scheltens P. MRI of the medial temporal lobe for the

diagnosis of alzheimer disease. In: Qizilbash N, Schneider LS,

Chui H, Tarriot P, Brodaty H, Kaye J, et al., eds. Evidence-Based De-

mentia Practice. Oxford, United Kingdom: Blackwell Science; 2002.

p. II. 4.7.

[6] Geuze E, Vermetten E, Bremner JD. MR-based in vivo hippocampal

volumetrics: 1. Review of methodologies currently employed. Mol

Psychiatry 2005;10:147e59.[7] Barnes J, Bartlett JW, van de Pol LA, Loy CT, Scahill RI, Frost C, et al.

A meta-analysis of hippocampal atrophy rates in Alzheimer’s disease.

Neurobiol Aging 2008;30:1711e23.[8] Colliot O, Chetelat G, Chupin M, Desgranges B, Magnin B, Benali H,

et al. Discrimination between Alzheimer disease, mild cognitive im-

pairment, and normal aging by using automated segmentation of the

hippocampus. Radiology 2008;248:194e201.

[9] Brewer JB, Magda S, Airriess C, Smith ME. Fully-automated

quantification of regional brain volumes for improved detection

of focal atrophy in Alzheimer disease. Am J Neuroradiol 2009;

30:578e80.[10] Morra JH, Tu Z, Apostolova LG, Green AE, Avedissian C,

Madsen SK, et al. Validation of a fully automated 3D hippocampal

segmentation method using subjects with Alzheimer’s disease mild

cognitive impairment, and elderly controls. Neuroimage 2008;

43:59e68.

[11] Barnes J, Foster J, Boyes RG, Pepple T, Moore EK, Schott JM, et al.

A comparison of methods for the automated calculation of volumes

and atrophy rates in the hippocampus. Neuroimage 2008;40:

1655e71.[12] Duchesne S, Pruessner J, Collins DL. Appearance-based segmentation

of medial temporal lobe structures. Neuroimage 2002;17:515e31.

[13] Frisoni GB, Boccardi M, Ganzola R, Duchesne S, Robitaille N, Re-

dolfi A, et al. Survey of protocols for manual hippocampal volumetry:

preparatory steps for an EADC-ADNI harmonized protocol. In: Pro-

ceedings of the American Academy of Neurology Annual Meeting;

April 10e17, 2010; Toronto, Canada.

[14] Watson C, Andermann F, Gloor P, Jones-Gotman M, Peters T,

Evans A, et al. Anatomic basis of amygdaloid and hippocampal vol-

ume measurement by magnetic resonance imaging. Neurology 1992;

42:1743e50.

[15] FoxNC, Black RS, Gilman S, RossorMN, Griffith SG, Jenkins L, et al.

Effects of Abeta immunization (AN1792) on MRI measures of cere-

bral volume in Alzheimer disease. Neurology 2005;64:1563e72.

[16] Insausti R, Juottonen K, Soininen H, Insausti AM, Partanen K,

Vainio P, et al. MR volumetric analysis of the human entorhinal, peri-

rhinal, and temporopolar cortices. Am JNeuroradiol 1998;19:659e71.

[17] Machulda MM, Ward HA, Cha R, O’Brien P, Jack CR Jr. Functional

inferences vary with the method of analysis in fMRI. Neuroimage

2001;14:1122e7.

[18] Sparks DL, Lemieux SK, Haut MW, Baxter LC, Johnson SC,

Sparks LM, et al. Hippocampal volume change in the Alzheimer dis-

ease cholesterol-lowering treatment trial. Cleve Clin J Med 2008;

75:S87e93.

[19] Freeborough PA, Fox NC. The boundary shift integral: an accurate and

robust measure of cerebral volume changes from registered repeat

MRI. IEEE Trans Med Imaging 1997;16:623e9.

[20] Jones RW, Kivipelto M, Feldman H, Sparks L, Doody R, Waters DD,

et al. The Atorvastatin/Donepezil in Alzheimer’s Disease Study

(LEADe): design and baseline characteristics. Alzheimers Dement

2008;4:145e53.

[21] Feldman HH, Doody RS, Kivipelto M, Sparks DL, Waters DD,

Jones RW, et al. Randomized controlled trial of atorvastatin in mild

to moderate Alzheimer disease: LEADe. Neurology 2010;74:956e64.[22] Jack CR Jr, Twomey CK, Zinsmeister AR, Sharbrough FW,

Petersen RC, Cascino GD. Anterior temporal lobes and hippocampal

formations: normative volumetric measurements for MR images in

young adults. Radiology 1989;172:549e54.

[23] Jack CR Jr, Shiung MM, Gunter JL, O’Brien PC, Weigand SD,

Knopman DS, et al. Comparison of different MRI brain atrophy rate

measures with clinical disease progression in AD. Neurology 2004;

62:591e600.

[24] Jack CR Jr, Petersen RC, GrundmanM, Jin S, Gamst A,Ward CP, et al.

LongitudinalMRI findings from the vitamin E and donepezil treatment

study for MCI. Neurobiol Aging 2008;29:1285e95.[25] Gauthier S, Aisen PS, Ferris SH, Saumier D, Duong A, Haine D, et al.

Effect of tramiprosate in patients with mild-to-moderate Alzheimer’s

disease: exploratory analyses of the MRI sub-group of the Alphase

study. J Nutr Health Aging 2009;13:550e7.[26] Vellas B, Andrieu S, Sampaio C, Coley N, Wilcock G, European Task

Force Group. Endpoints for trials in Alzheimer’s disease: a European

task force consensus. Lancet Neurol 2008;7:436e50.

[27] Zipursky RB, Marsh L, Lim KO, DeMent S, Shear PK, Sullivan EV,

et al. Volumetric MRI assessment of temporal lobe structures in

schizophrenia. Biol Psychiatry 1994;35:501e16.