AAN –Toronto, April 11, 2010 Development of a harmonized protocol for hippocampal tracing An EADC-ADNI joint effort INTRODUCTION

Structure: MR hippo volumetry Biochemistry: CSF tau/Abeta Function: FDG PET Amyloid deposits: PET BACKGROUND Hippocampal volume as a diagnostic marker Dubois, Feldman et al., Lancet Neurol 2007

Geuze et al., Mol Psychiatry 2005;10: Ref.Med border Lat border Inf border Norm. hippo vol (cm 3 ) LeftRight Watson et al. Mesial edge of temporal lobe Temp horn of lat ventr Incl subicular complex & uncal cleft w/ border separating subicular complex from parahippo gyrus Zipursky et al. Regional outline at choroidal fissure Not mentioned The interface of hippocampal tissue and parahippocampal gyrus white matter BACKGROUND The effect of segmentation protocols on hippocampal volume

BACKGROUND Hippocampal volume as an outcome measure in trials of disease modifiers DrugStudyEffectSegment. methodRef TramiprosateAlphase 68% (100 mg) 120% (150 mg) Not mentionedGauthier et al., JNHA 2009 AtorvastatinADCLT “CHV* mm³ -134±174 vs - 583±354 p>.05” Manual protocol Insausti et al., 1998 AJNR; Machulda et al., 2001, Neuroimage Sparks et.a,l CCJM,2008 LEADe “significant at p<.05” MIDAS (semiautomated) Jones et al., Alz&Dem 2008 Feldman et al., Neurology 2010 AN1792 “CHV* % 3.78±2.63 vs 2.86±3.19 p = 0.124” Manual protocol Watson et al., 1992, Neurology Fox N et al., Neurology, 2005 Xaliproden Sanofi- Aventis Trials “significant atrophy” §Not mentionedFrom Vellas B et al. (Review), Lancet Neurol.,2008 Donepezil “CHV* % -6.14±3.49vs ±2.28 p= 0.07” ε4 carriers only Manual protocol Jack et al., 1989;Radiology Jack et al., 2004; Neurology Jack CR et al., Neurobiol. Aging, 2008 * CHV= Change in hippocampal volume; § referenced link not working

BACKGROUND Hippocampal volume (manual segmentation) as gold standard for automated segmentation algorithms Morey et al., NeuroImage 2009

To Harmonize Available Protocols for Manual Hippocampal Segmentation To Test the Validity of the Harmonized Protocol AIMS

F Barkhof/P ScheltensAmsterdam B Dubois/S LehericiParis N FoxLondon GB Frisoni Brescia H Hampel/J PantelFrankfurt A SimmonsLondon H SoininenKuopio S TeipelRostock LO WahlundStockholm THE WORKGROUP EADC

CR Jack Jr Rochester, MN G Barzokis UCLA, CA D Bennet Chicago, IL J Csernansky NorthW U, WA C DeCarli UC Davis, CA M De Leon New York, NY J Kaye Portland, OR R Killiany Boston USM, MA PM Thompson LoNI, UCLA, CA M Weiner/S Mueller UCSF/VAMC THE WORKGROUP ADNI

THE WORKGROUP other relevant hippo experts R CamicioliUniv Alberta, Canada J O’BrienNewCastle, UK J PruessnerMcGill, QC, Canada

EADC-ADNI P.I.s B VellasToulouse, France B WinbladStockholm, Sweden M WeinerUCSF, US Population-based studies M Breteler/ T Den HeijerRotterdam Scan Study P Sachdev/ JJ Maller PATH through life L LaunerNIA Bethesda W JagustBerkeley THE WORKGROUP Advisory Boards

Statistical Board S DuchesneLaval Univ, Canada L CollinsMNI, McGill, Montreal P PasqualettiAFaR, Rome Clinical Advisor PJ VisserMaastricht, The Netherlands Dissemination/Education G WaldemarCopenhagen, Denmark THE WORKGROUP Boards

Marina Boccardi psychologist, PhD Rossana Ganzola psychologist Enrica Cavedopsychologist Michela Pievanimathematician Anna Carolimathematician Alberto Redolfibioinformaticist THE WORKGROUP Personnel in Brescia

AAN –Toronto, April 11, 2010 Development of a harmonized protocol for hippocampal tracing An EADC-ADNI joint effort METHODS

Operationalization of protocols’ differences and extraction of tracing units Quantification of tracing units’ features Harmonized protocol METHODS

Selection criteria i)Being based on 3D T1 MRI with field strength greater than 1 Tesla ii)Providing explicit description of landmarks for tracing of hippocampal borders iii)Being validated on AD/MCI samples, or being widely adopted in the literature about AD. OPERATIONALIZATION Selection of segmentation protocols

Plane of tracing: AC-PC line; Normalization to the Talairach space Tracing start from the tail of hippocampus to the hippocampal head In the following slices, you can find the hippocampal tracing on consecutive coronal slices (1.5T) of the control subject and AD patient according to Pruessner's criteria OPERATIONALIZATION Protocols’ features extraction

Tracings carried out on a control and an AD subject from the ADNI dataset Tracings sent to the Authors for correction or approval OPERATIONALIZATION Prototypical tracings

CTRL Most posterior slice : slice where an ovoid mass of gray matter (1) started to appear inferomedially of the atrium of lateral ventricle (2) Sagittal view 2= Atrium of lateral ventricle 4= Parahippocampal gyrus 5= Gyrus dentatus 6= Cornu Ammonis 7= Gyrus fasciolaris 8= Gyrus of Andreas-Retzius 9= Isthmus The fornix can be excluded by a horizontal line (in fuchsia) from the superior border of the quadrigeminal cistern (3) to the lateral ventricle (2). Part of gyrus fasciolaris (7) and of Andreas-Retzius gyrus (8) can be excluded by a vertical line (in red) from the medial end of the lateral ventricle (2) down to the parahippocampal gyrus (4).

AD Most posterior slice : slice where an ovoid mass of gray matter (1) started to appear inferomedially of the atrium of lateral ventricle (2) 2= Atrium of lateral ventricle 4= Parahippocampal gyrus 5= Gyrus dentatus 6= Cornu Ammonis 7= Gyrus fasciolaris 8= Gyrus of Andreas Retzius 9= Isthmus Sagittal view In AD subjects the cerebral structures are atrophic, lateral ventricle (2) and ambient cistern (3) result enlarged. Therefore, the lines used to exclude the gyrus fasciolaris, the Andreas- Retzius gyrus and the fornix do not intersect the hippocampal tail. That being so, is there another way to exclude the gyrus fasciolaris, the Andreas-Retzius gyrus and the fornix in these subjects?

For each protocol: features extraction, tracing and author’s certification AD CTRL X 10 OPERATIONALIZATION Authors’ check

OPERATIONALIZATION Extraction of similarities and differences Harmonized language Reduction of redundancy Tracing units and subunits

OPERATIONALIZATION Extraction of similarities and differences

Plane of tracing Axis of hippocampus [B,C,J,L,S] AC-PC line [H,K,M,Pa,Pr] Most posterior slice Where inferior and superior colliculi are jointly visualized [B] Where crus of fornix is visible in full profile [C,K,L] Where the crura of the fornices are seen in full profile on both sides [J,S] Where gray matter is visible inferomedially to the trigone of the lateral ventricle [H,M,Pa,Pr] Superior border Lower border of alveus/fimbria [B,H,K,Pa,S] Upper border of alveus/fimbria [C,J,L,M,Pr] Horizontal line from the superior border of the quadrigeminal cistern to the lateral ventricle (tail) [Pr] Separation subiculum/enthorinal cortex vertical line from the CA to the WM of the parahippocampal gyrus [C] 45° line connecting the most inferior part of the subiculum medially to the cistern (head and body) [Pr] Oblique line with same inclination of parahippocampal WM, connecting the inferior part of the subiculum to the quadrigeminal cistern [K,L,M ¥ ] Orizontal line from the highest medial point of the parahippocampal WM to the cistern [B,H,J ¥ ] Line outlining the contour of white matter of parahippocampal gyrus [Pa,S]

Operationalization of protocols’ differences Quantification of tracing units’ features Harmonized protocol METHODS

Contribution to hippo volume in control and AD Effect on re-test variability Contribution to differences between AD and controls QUANTIFICATION OF TRACING UNITS’ FEATURES

Operationalization of protocols’ differences Quantification of tracing units’ features Harmonized protocol METHODS

Participants to a Delphi panel based on: expertise in hippocampal anatomy manual tracing scientific production Assisted decisions based on: Low variability, highly informative....included in harmonized protocol High variability, little informative....excluded from harmonized prot HARMONIZED PROTOCOL

Validation with neuropathological data Comparison with currently used protocols Public tracings and probability maps Standard environment for tracing, learning, and certification HARMONIZED PROTOCOL Validation & implementation

AAN –Toronto, April 11, 2010 Development of a harmonized protocol for hippocampal tracing An EADC-ADNI joint effort EARLY RESULTS

Protocols’ features extraction & differences operationalization Modeling & computations of the operationalized differences Discussion EARLY RESULTS

B Bartzokis G, et al. Psychiatry Res 1998;82:11-24 C Convit A, et al. Neurobiol Aging 1997;18:131-8 HHaller JW, et al. Radiology 1997;202: JJack CR Jr. Epilepsia 1994;35 - Suppl 6:S21-9 KKilliany RJ, et al. Arch Neurol 1993;50: LLehéricy S, et al. Am J Neuroradiol 1994;15: MMalykhin NV, et al. Psychiatry Res 2007;155: PaPantel J, et al. Hippocampus 2000;10:752-8 PrPruessner JC, et al. Cereb Cortex 2000;10: SSoininen HS, et al. Neurology 1994;44: PROTOCOLS’ FEATURES EXTRACTION

PROTOCOLS’ FEATURES EXTRACTION

Harmonized language Reduction of redundancy Tracing units and subunits

Plane of tracing Axis of hippocampus [B,C,J,L,S] AC-PC line [H,K,M,Pa,Pr] Most posterior slice Where inferior and superior colliculi are jointly visualized [B] Where crus of fornix is visible in full profile [C,K,L] Where the crura of the fornices are seen in full profile on both sides [J,S] Where gray matter is visible inferomedially to the trigone of the lateral ventricle [H,M,Pa,Pr] Superior border Lower border of alveus/fimbria [B,H,K,Pa,S] Upper border of alveus/fimbria [C,J,L,M,Pr] Horizontal line from the superior border of the quadrigeminal cistern to the lateral ventricle (tail) [Pr] Separation subiculum/enthorinal cortex vertical line from the CA to the WM of the parahippocampal gyrus [C] 45° line connecting the most inferior part of the subiculum medially to the cistern (head and body) [Pr] Oblique line with same inclination of parahippocampal WM, connecting the inferior part of the subiculum to the quadrigeminal cistern [K,L,M ¥ ] Orizontal line from the highest medial point of the parahippocampal WM to the cistern [B,H,J ¥ ] Line outlining the contour of white matter of parahippocamp al gyrus [Pa,S]

No longer an issue thanks to software for 3D navigation PROTOCOLS’ FEATURES EXTRACTION Most anterior slice

H. Haller-Csernansky et al 1997 K. Killiany et al 1993 M. Malykhin et al 2007 Pa. Pantel et al 2000 Pr. Pruessner et al 2000 B. Bartzokis et al 1998 C. Convit et al 1997 J. Jack 1994 L. Lehéricy et al 1994 S. Soininen H et al 1994 Axis of hippocampusAC-PC PROTOCOLS’ FEATURES EXTRACTION Plane of tracing

PROTOCOLS’ FEATURES EXTRACTION Most posterior slice

A. Level where the inferior (1) and superior (2) colliculi are jointly visualized [B] B. Slice where the crus of fornix is visible in full profile (3) [C, K, L] C. Slice where the crura of fornices on both sides are seen in full profile (4, 5) [J, S] D. Slice where an ovoid mass of gray matter (6) appears inferomedially to the trigone of the lateral ventricle (posterior to anterior) [H, M, Pa, Pr]

Yellow = Alveus PROTOCOLS’ FEATURES EXTRACTION Alveus

Yellow = Fimbria PROTOCOLS’ FEATURES EXTRACTION Fimbria

PROTOCOLS’ FEATURES EXTRACTION Upper border

PROTOCOLS’ FEATURES EXTRACTION Medial border Vertical line [C] 45° line [Pr] Line with same inclination WM [K, L, M] Horizontal line [B, H, J] Morphological details [Pa, S]

EARLY RESULTS Protocols’ features extraction & differences operationalization Modeling & computations of the operationalized differences Discussion

Plane of tracing Axis of hippocampus [B,C,J,L,S] AC-PC line [H,K,M,Pa,Pr] Most posterior slice Where inferior and superior colliculi are jointly visualized [B] Where crus of fornix is visible in full profile [C,K,L] Where the crura of the fornices are seen in full profile on both sides [J,S] Where gray matter is visible inferomedially to the trigone of the lateral ventricle [H,M,Pa,Pr] Superior border Lower border of alveus/fimbria [B,H,K,Pa,S] Upper border of alveus/fimbria [C,J,L,M,Pr] Separation subiculum/enthorinal cortex vertical line from the CA to the WM of the parahippocampal gyrus [C] 45° line connecting the most inferior part of the subiculum medially to the cistern (head and body) [Pr] Oblique line with same inclination of parahippocampal WM, connecting the inferior part of the subiculum to the quadrigeminal cistern [K,L,M ¥ ] Orizontal line from the highest medial point of the parahippocampal WM to the cistern [B,H,J ¥ ] Line outlining the contour of white matter of parahippocampal gyrus [Pa,S] MODELING & COMPUTATIONS

Modeling by Simon Duchesne and Nicolas Robitaille Université Laval and Centre de Recherche Université Laval – Robert Giffard Quebec City, Canada

MODELING & COMPUTATIONS

J Pr M PaH BCK L S MODELING & COMPUTATIONS

DISCUSSION Features operationalization: 4 tracing units, and subunits Preliminary info on contribution of units to total hippo volume 3D models may help Delphi panel for developing harmonized protocol Available: draft paper, and online material

April 14: Poster 24 PO3: Ageing and dementia imaging - other

Lilly International Wyeth International ACKNOWLEDGEMENTS