Tau PET burden in Brodmann areas 35 and 36 is associated with individual differences in cognition in non-demented older adults

Abstract

Introduction
The accumulation of neurofibrillary tau tangles, a neuropathological hallmark of Alzheimer’s disease (AD), occurs in medial temporal lobe (MTL) regions early in the disease process, with some of the earliest deposits localized to subregions of the entorhinal cortex. Although functional specialization of entorhinal cortex subregions has been reported, few studies have considered functional associations with localized tau accumulation.

Methods
In this study, stepwise linear regressions were used to examine the contributions of regional tau burden in specific MTL subregions, as measured by ¹⁸ F-MK6240 PET, to individual variability in cognition. Dependent measures of interest included the Clinical Dementia Rating Sum of Boxes (CDR-SB), Mini Mental State Examination (MMSE), and composite scores of delayed episodic memory and language. Other model variables included age, sex, education, APOE4 status, and global amyloid burden, indexed by ¹¹ C-PiB.

Results
Tau burden in right Brodmann area 35 (BA35), left and right Brodmann area 36 (BA36), and age each uniquely contributed to the proportion of explained variance in CDR-SB scores, while right BA36 and age were also significant predictors of MMSE scores, and right BA36 was significantly associated with delayed episodic memory performance. Tau burden in both left and right BA36, along with education, uniquely contributed to the proportion of explained variance in language composite scores. Importantly, the addition of more inclusive ROIs, encompassing less granular segmentation of the entorhinal cortex, did not significantly contribute to explained variance in cognition across any of the models.

Discussion
These findings suggest that the ability to quantify tau burden in more refined MTL subregions may better account for individual differences in cognition, which may improve the identification of non-demented older adults who are on a trajectory of decline due to AD.

Full text

Frontiers in Aging Neuroscience 01 frontiersin.org
Tau PET burden in Brodmann
areas 35 and 36 is associated with
individual dierences in cognition
in non-demented older adults
NishaRani
1†, KylieH.Alm
1†, Caitlin A.Corona-Long
1,
CarolineL.Speck
1, AnjaSoldan
2, CorinnePettigrew
2, YuxinZhu
2,
MarilynAlbert
2 and ArnoldBakker
1,2*
1 Department of Psychiatry and Behavioral Sciences, Johns Hopkins University School of Medicine,
Baltimore, MD, United States, 2 Department of Neurology, Johns Hopkins University School of Medicine,
Baltimore, MD, United States
Introduction: The accumulation of neurofibrillary tau tangles, a neuropathological
hallmark of Alzheimer’s disease (AD), occurs in medial temporal lobe (MTL)
regions early in the disease process, with some of the earliest deposits localized
to subregions of the entorhinal cortex. Although functional specialization of
entorhinal cortex subregions has been reported, few studies have considered
functional associations with localized tau accumulation.
Methods: In this study, stepwise linear regressions were used to examine the
contributions of regional tau burden in specific MTL subregions, as measured
by 18F-MK6240 PET, to individual variability in cognition. Dependent measures
of interest included the Clinical Dementia Rating Sum of Boxes (CDR-SB), Mini
Mental State Examination (MMSE), and composite scores of delayed episodic
memory and language. Other model variables included age, sex, education,
APOE4 status, and global amyloid burden, indexed by 11C-PiB.
Results: Tau burden in right Brodmann area 35 (BA35), left and right Brodmann
area 36 (BA36), and age each uniquely contributed to the proportion of explained
variance in CDR-SB scores, while right BA36 and age were also significant
predictors of MMSE scores, and right BA36 was significantly associated with
delayed episodic memory performance. Tau burden in both left and right BA36,
along with education, uniquely contributed to the proportion of explained
variance in language composite scores. Importantly, the addition of more
inclusive ROIs, encompassing less granular segmentation of the entorhinal
cortex, did not significantly contribute to explained variance in cognition across
any of the models.
Discussion: These findings suggest that the ability to quantify tau burden in more
refined MTL subregions may better account for individual dierences in cognition,
which may improve the identification of non-demented older adults who are on a
trajectory of decline due to AD.
KEYWORDS
18F-MK6240 Tau PET, entorhinal cortex, mild cognitive impairment, episodic memory,
BA35, BA36
OPEN ACCESS
EDITED BY
P. Hemachandra Reddy,
Texas Tech University Health Sciences Center,
UnitedStates
REVIEWED BY
Gabriel Gonzalez-Escamilla,
Johannes Gutenberg University Mainz,
Germany
Manisha Thaker,
Scintillon Institute, UnitedStates
Elisa De Paula Franca Resende,
Federal University of Minas Gerais, Brazil
*CORRESPONDENCE
Arnold Bakker
abakker@jhu.edu
†These authors have contributed equally to this
work and share first authorship
RECEIVED 04 August 2023
ACCEPTED 23 October 2023
PUBLISHED 14 December 2023
CITATION
Rani N, Alm KH, Corona-Long CA, Speck CL,
Soldan A, Pettigrew C, Zhu Y, Albert M and
Bakker A (2023) Tau PET burden in Brodmann
areas 35 and 36 is associated with individual
dierences in cognition in non-demented
older adults.
Front. Aging Neurosci. 15:1272946.
doi: 10.3389/fnagi.2023.1272946
COPYRIGHT
© 2023 Rani, Alm, Corona-Long, Speck,
Soldan, Pettigrew, Zhu, Albert and Bakker. This
is an open-access article distributed under the
terms of the Creative Commons Attribution
License (CC BY). The use, distribution or
reproduction in other forums is permitted,
provided the original author(s) and the
copyright owner(s) are credited and that the
original publication in this journal is cited, in
accordance with accepted academic practice.
No use, distribution or reproduction is
permitted which does not comply with these
terms.
TYPE Original Research
PUBLISHED 14 December 2023
DOI 10.3389/fnagi.2023.1272946

Rani et al. 10.3389/fnagi.2023.1272946
Frontiers in Aging Neuroscience 02 frontiersin.org
1 Introduction
ere is considerable evidence that the pathological hallmarks of
Alzheimer’s disease (AD) emerge many years before clinical symptoms
of mild cognitive impairment (MCI; Sperling etal., 2011; Pletnikova
etal., 2018). Amyloid (Aβ) plaques and neurobrillary tau tangles, the
neuropathological hallmarks of AD, have been shown to accumulate
among cognitively normal (CN) older adults, with the percentage of CN
individuals showing these pathological changes varying with age (Brier
etal., 2016). Biomarker evidence of these pathological changes among
asymptomatic individuals is reected in studies involving the assessment
of cerebrospinal uid (CSF; Blennow etal., 2010) and positron emission
tomography (PET; Bao et al., 2017). Importantly, PET studies can
provide details about the spatial distribution of amyloid and tau in the
brain during the early phases of AD.
While the distribution of amyloid deposition in the brain early
in the course of disease does not appear to be particularly
informative about the likelihood of progression from normal
cognition to MCI (Scheinin etal., 2009), tau tends to deposit in a
systematic fashion early in the course of AD and then spreads in a
relatively predictable manner to other brain regions (Pascoal etal.,
2018). Post-mortem studies have provided valuable insights into the
spatial and temporal progression of tau pathology, revealing that the
formation of neurobrillary tangles in the initial stages occurs in the
entorhinal cortex (EC), specically in the transentorhinal cortex
(TEC), which serves as the transition between the lateral portions of
the EC and the perirhinal cortex (Braak and Braak, 1990; Braak
etal., 2006; Kaufman etal., 2018). Notably, tau accumulation in the
TEC is common by age 60, even among CN older adults and in the
absence of concurrent Aβ pathology (Maass et al., 2018). Tau
pathology then continues to spread through other medial temporal
lobe (MTL) regions (Braak etal., 2006), which are of critical interest
given their well-established roles in episodic memory functioning
(Dickerson and Eichenbaum, 2010). Moreover, tau deposition has
been observed to exhibit a more consistent and robust association
with cognitive decline throughout aging and the AD spectrum
(Nelson etal., 2012; Brier etal., 2016; Ossenkoppele etal., 2020),
compared to amyloid deposition.
Fortunately, tau-specic PET tracers permit the investigation of
regional associations between tau pathology and cognition in vivo
(Johnson etal., 2016; Ossenkoppele etal., 2016, 2020; Chiotis etal.,
2021). Over the past decade, numerous PET radiotracers have been
introduced to visualize tau pathology deposition in vivo (Ossenkoppele
et al., 2016, 2021; Hall etal., 2017; Ricci etal., 2021). Compared to
initial tracers,
18
F-MK6240 (Malarte etal., 2021), a second-generation
tau PET tracer, has demonstrated favorable imaging characteristics
and spatial distributions consistent with the spread of neurobrillary
tangles reported in autopsy tissue (Pascoal etal., 2018; Lohith etal.,
2019; Betthauser et al., 2020). Importantly,
18
F-MK6240 exhibits
subnanomolar anity for tau tangles, with a dissociation constant of
approximately 0.3 nM, making it an improved tau radioligand, with
minimal binding to o-target sites in the basal ganglia and choroid
plexus (Hostetler etal., 2016). Comparative studies with
18
F-AV1451
PET have shown that 18F-MK6240 oers a higher dynamic range of
standardized uptake value ratio (SUVR) values across dierent Braak
stages in AD patients, indicating its potential for enhanced sensitivity
for early detection and monitoring of AD progression (Gogola etal.,
2022; Krishnadas etal., 2023).
e high binding specicity of
18
F-MK6240 PET imaging provides
an opportunity to investigate the relationship between tau burden in
specic MTL subregions and individual dierences in cognitive
functioning among older adults. For example, recent studies in rodents
have shown that the lateral EC and perirhinal cortex support encoding
of object/content information, while medial EC and parahippocampal
regions facilitate encoding of context and spatial information (Knierim
et al., 2014). Functional specialization in these regions also appears
dierentially impacted by aging and AD, with more lateral EC-dependent
functions tending to bepreferentially impacted by aging (Reagh etal.,
2018; Tran etal., 2021, 2022). Since both AD-related tau pathology and
age-related functional changes relate to MTL structures non-uniformly,
subregion measurements may bemore sensitive to early disease changes,
compared to currently employed segmentation approaches.
erefore, the primary objective of this study was to examine the
associations between tau accumulation in MTL subregions and
individual dierences in cognition in a cohort of non-demented older
adults. Utilizing detailed segmentation of MTL subregions with
advanced registration approaches, this study examined whether
quantifying tau burden in smaller MTL subregions accounts for a
greater proportion of the variance in cognitive performance when
compared to larger MTL areas utilized by traditional PET image
analysis approaches. A total of 8 regions of interest (ROIs) were selected
a priori based on the location of post-mortem tau accumulation in the
early stage of AD (Braak and Braak, 1990; Braak etal., 2006). ese
regions included the le and right EC obtained from FreeSurfer
parcellations (Fischl, 2012), as well as the le and right entorhinal
cortex (labeled ERC to distinguish the dierent sowares), Brodmann
area 35 (BA35), and Brodmann area 36 (BA36) obtained from the
Automated Segmentation of Hipppocampal Subelds (ASHS) soware
(Xie etal., 2019). FreeSurfer EC parcellations encompass the anterior
portion of the parahippocampal gyrus, including the medial bank of
the collateral sulcus, as well as a portion of the TEC (Braak and Braak,
1991; Taylor and Probst, 2008). ASHS BA35 largely overlaps with TEC
(Braak and Braak, 1991; Braak et al., 2006) and a portion of the
perirhinal cortex, while BA36 predominantly encompasses the
perirhinal cortex (Xie etal., 2019). ese segmentations were applied
to imaging data obtained from the Biomarkers of Cognitive Decline
Among Normal Individuals (BIOCARD) study that has detailed
clinical and cognitive evaluations of the participants, as well as
extensive biomarker data (Albert etal., 2014).
2 Methods
2.1 Participants
e present investigation utilized data from a subset of
participants enrolled in the BIOCARD study, which is an ongoing
longitudinal prospective cohort study aimed at understanding the
early phases of AD. e BIOCARD study began in 1995 at the
intramural program of the National Institutes of Health (NIH) and
continued at Johns Hopkins University (JHU) starting in 2009. All
participants were cognitively normal when enrolled and, by design,
75% had a family history of dementia. When the study was conducted
at the NIH, clinical and cognitive assessments were completed
annually and CSF, blood, and magnetic resonance imaging (MRI)
scans were collected approximately every other year. In 2015, biennial

Rani et al. 10.3389/fnagi.2023.1272946
Frontiers in Aging Neuroscience 03 frontiersin.org
collection of MRI, CSF, and amyloid PET data commenced, and tau
PET imaging data collection using
18
F-MK6240 began in 2020. e
participants were primarily middle-aged (M = 57.3, SD = 10.4,
range = 20.0–85.8) at enrollment. For further information on the
BIOCARD cohort, see Albert etal. (2014).
e current study sample included all participants from the
BIOCARD study who underwent both a
18
F-MK6240 tau PET scan
and an amyloid PET scan using 11C-PiB. All included individuals also
had a concurrent T1-MPRAGE scan and cognitive testing obtained
during the same visit. e cognitive status of each participant in the
study sample was determined by a consensus diagnosis from the JHU
BIOCARD Clinical Core sta following the study visit. e diagnostic
approach adheres to the guidelines of the National Institute on Aging
– Alzheimer’s Association working group and is comparable to the
approach used at the National Institute on Aging Alzheimer’s Disease
Centers program (Albert etal., 2011; McKhann etal., 2011). First, a
syndromic diagnosis is established based on three sources of
information, including (1) clinical data on the individual’s medical,
neurological, and psychiatric status; (2) reports of cognitive changes
from the participant and their informants; and (3) evidence of
cognitive performance decline based on review of longitudinal
neuropsychological assessments of multiple cognitive domains with
comparison to published norms. Next, for participants with cognitive
impairment, the likely syndromic etiology was determined based on
the available neurologic, medical, and psychiatric information; more
than one etiology could beendorsed. All diagnoses were made without
knowledge of the imaging or uid biomarker data.
e resulting study sample consisted of 93 participants, including
82 CN individuals and 11 participants with a diagnosis of MCI.
2.2 Clinical and cognitive assessments
e clinical assessment of the participants included the
administration of the Clinical Dementia Rating Scale (CDR; Morris,
1993). e cognitive assessment included a battery of standardized
neuropsychological tests covering various cognitive domains,
including memory, executive function, language, visuospatial ability,
attention, speed of processing, and psychomotor speed. A detailed
description of these tests has been previously published (Albert etal.,
2014). Our analysis focused on several key measures, including the
Clinical Dementia Rating Sum of Boxes (CDR-SB) score, the Mini-
Mental State Examination (MMSE) score, as well as composite scores
for delayed episodic memory and language. Briey, the language
composite score was established using conrmatory factor analysis
(Soldan etal., 2019) and was based on three tests: Boston Naming
Test, Category Fluency (animals), and Letter Fluency (FAS). For each
of these tests, scores were z-transformed and weighted by their
respective standardized factor loadings from a conrmatory factor
analysis. e resulting transformed scores were averaged to obtain
language composite scores for each subject [see Soldan etal. (2019)
for further details]. e delayed episodic memory composite score
[previously described by Alm etal. (2022)] was based on the California
Verbal Learning Test (CVLT) long delay free recall and the Wechsler
Memory Scale Logical Memory (LM) delayed recall. Task scores were
z-transformed, giving equal weight to each memory measure, and
transformed scores were then averaged to obtain delayed episodic
memory composite scores for each subject.
2.3 MRI imaging
T1-weighted MRI images were acquired using a magnetization-
prepared rapid acquisition with gradient echo (MPRAGE) sequence
on a 3 T MRI scanner with a 32-channel head coil (Philips Achieva,
Eindhoven, Netherlands) to establish anatomical reference for PET
image registration. e following parameters were employed:
repetition time (TR)/echo time (TE) = 6.8 ms/3.1 ms, shot
interval = 3,000 ms, inversion time = 843 ms, ip angle = 8°, eld of
view (FOV) = 256 mm × 256 mm with 1.0 × 1.0 × 1.2 mm
3
voxels and
170 sagittal slices. Prior to image processing, a visual quality control
assessment was conducted to ensure the integrity and suitability of the
acquired images.
2.4 Amyloid and tau PET imaging
PET scans were acquired using a GE DISCOVERY RX PET/
computed tomography (CT) scanner in 3D acquisition mode. Tau
tangle 18F-MK6240 PET images were acquired at 90 min ± 1 min aer
a single bolus intravenous injection of 5.0 mCi ± 10% mCi
(volume ≤ 10 mL) of the radiotracer followed by a 10 to 20 mL saline
ush.
11
C-PiB was used for amyloid PET imaging, and images were
acquired in a 20-min brain scan session 50 min ± 1 min aer injection
of 15 mCi ± 1.5 mCi of radiotracer. Syringes were used to measure the
residual activity post-injection.
e PET images were reconstructed into six frames for
18
F-
MK6240 and four frames for
11
C-PiB using the three-dimensional
ordinary Poisson ordered-subset expectation maximization (3D
OP-OSEM) algorithm with corrections applied for detector eciency,
decay, dead time, attenuation, and scatter (Hudson and Larkin, 1994).
Each PET data frame was evaluated to verify adequate count statistics
and absence of head motion. Low-dose CT (120 KeV, 80 mA; Slice
thickness 3.75 mm; Slice separation 3.3 mm; FOV 500 mm) based
attenuation correction (STANDARD Kernel) was applied. e nal
PET images were reconstructed into 2 × 2 × 3.27 mm3 voxels in units
of radioactivity concentrations (Bq/ml).
To generate Standard Uptake Value (SUV, g/mL) PET images, the
Bq/ml PET images were normalized to the patient’s body weight
(BWt) and the injected dose (ID). e SUVs were calculated as the
mean radioactivity per injected dose per weight using the formula
SUV = A/(ID x BWt), where A represents the activity concentration of
the PET image in Bq/mL, BWt is the patient’s body weight in grams,
and ID is the injected dose in Bq. To account for radioactive decay, all
SUVs were corrected based on the specic half-life of the F-18
radionuclide (for
18
F-MK6240 PET) and C-11 radionuclide (for
11
C-
PiB PET). Participants were instructed to remain still for the total
duration of each PET scan.
11
C-PiB PET data were missing for 2 CN
participants. Representative
18
F-MK6240 PET image examples are
shown in Supplementary Figure S1.
2.5 PET image analysis
Both
18
F-MK6240 and
11
C-PiB PET images were motion-corrected
using rigid body linear registration with six degrees of freedom
implemented in FSL [FMRIB (Oxford Centre for Functional MRI of
the Brain) Soware Library (Jenkinson etal., 2002)]. A mean PET

Rani et al. 10.3389/fnagi.2023.1272946
Frontiers in Aging Neuroscience 04 frontiersin.org
image was calculated for each subject from the motion-corrected time
series, and registered and resampled to the subject’s skull-stripped
T1-weighted image using FreeSurfer (Dale etal., 1999; Greve and
Fischl, 2009). To correct for partial volume eects (PVE) resulting
from the limited spatial resolution of the PET scan, partial volume
correction (PVC) was performed using the geometric transfer matrix
(GTM) method using tools integrated into PETSurfer (Rousset etal.,
1998). e GTM method assumes that within a specic ROI, the tissue
is homogeneous. Because age-related changes may aect tissue
homogeneity, particularly within our ROIs, the GTM method was
combined with the region-based voxel-wise (RBV) approach, enabling
voxel-wise PVE corrections (omas etal., 2011; Shidahara etal.,
2017). e RBV correction relies on anatomical parcellation and an
accurate point-spread function (PSF) estimation. In our PET data
analysis, weemployed a gaussian kernel with an isotropic full-width
at half-maximum (FWHM) of 4.0 mm for the GTM-RBV-based PVC,
which accounted for spill-over eects between dierent compartments
during voxel-based correction. e T1-weighted MRI images were
warped into Montreal Neurological Institute (MNI) space using the
Advanced Normalization Tools (ANTS) soware package (Avants
etal., 2014; Tustison etal., 2014). e resulting transformation matrix
from this registration was applied to each subject’s PVC-corrected
PET images for normalization to MNI space. A diagram summarizing
PET image analysis steps is displayed in Supplementary Figure S2.
18
F-MK6240 PET images underwent intensity normalization by
utilizing the average uptake in the pons derived from the FreeSurfer
parcellation as the reference region. e choice of pons as a reference
region was based on its lower variability and lower extracerebral
contamination in CN and cognitively impaired subjects (Fu etal.,
2023). e
18
F-MK6240 SUVR PET images in MNI template space
were utilized to generate mean SUVR values to quantify tau
accumulation in subregions within the MTL. e mean 11C-PiB PET
SUV images were converted into SUVR images using whole
cerebellum gray matter dened by the FreeSurfer-derived MRI
parcellations eroded by 3-voxels in 3D as the reference region. e
choice of cerebellar gray matter as the reference region for
11
C-PiB was
based on its known minimal susceptibility to brillar amyloid
deposition in AD, making it a reliable estimator of nonspecic PiB
binding (Klunk etal., 2004; Price etal., 2005; Liu etal., 2015). e
11
C-PiB SUVR PET images in MNI template space were used to
generate mean SUVR values for amyloid-β deposition.
2.6 Image segmentation
e FreeSurfer image analysis pipeline (version 7.2.0) was
employed for cortical reconstruction and volumetric segmentation of
MPRAGE images, applying intensity normalization, skull stripping,
cortical surface extraction, volumetric segmentation, and surface-
based registration. Following reconstruction, the datasets were
visually reviewed for segmentation accuracy and any errors were
corrected (Dale etal., 1999; Fischl etal., 1999, 2002; Ségonne etal.,
2007; Fischl, 2012).
Segmentation of additional MTL subregions was subsequently
completed using the ASHS soware, using the ASHS-T1 atlas
specically designed for older adults (also known as ASHS-PMC-T1
atlas; Yushkevich etal., 2014, 2016; Xie etal., 2019). Briey, ASHS-T1
employs a series of steps to accomplish the automatic segmentation
task. First, the target MRI scan is up-sampled to a resolution of
0.5×0.5×1.0 mm3 using a non-local-mean super-resolution algorithm
(Manjon et al., 2010). en, symmetric greedy dieomorphic
registration within the ANTs soware (Avants etal., 2008) is used to
warp each segmentation atlas to the target MRI scan. A joint label
fusion algorithm combines the anatomical labels from the warped
atlases, generating a consensus segmentation. is fusion process
assigns spatially varying weights to each atlas based on patch-level
similarity to the target image, while considering potential redundancy
among the atlases (Wang et al., 2013). Additionally, a corrective
learning algorithm corrects systematic segmentation biases using
classiers trained from leave-one-out segmentation of the atlas images
(Wang etal., 2011). Bootstrapping is applied, leveraging the results of
multi-atlas segmentation to improve the matching between the atlas
and target images. e ASHS-generated ROIs were subsequently
down-sampled using FSL FLIRT (Jenkinson and Smith, 2001;
Jenkinson etal., 2002) to match the original MNI image resolution of
1.0×1.0×1.0 mm
3
. resholding and binarizing were accomplished
using fslmaths, using a threshold >0.9. is down-sampling facilitated
the extraction of SUVR measurements from the MNI-resampled tau
PET image within the created MTL ROIs.
2.7 SUVR computation
Mean SUVR for 18F-MK6240 was computed in eight ROIs, which
included the le and right EC obtained from FreeSurfer parcellations,
as well as the le and right ERC, BA35, and BA36 regions derived
from ASHS segmentations (Figure1). e FreeSurfer-dened EC
encompasses the anterior portion of the parahippocampal gyrus,
including the medial bank of the collateral sulcus, which may also
include the transentorhinal region (Braak and Braak, 1991; Taylor and
Probst, 2008) corresponding to BA35 or to the medial perirhinal
cortex. e FreeSurfer-derived label was obtained from each
participant’s native space T1-MPRAGE image and projected to the tau
PET images in MNI space. e ASHS labels included ERC, BA35, and
BA36, applied to the MNI-transformed
18
F-MK6240 PET data. e
utilization of the MNI common template space facilitated a direct
comparison of parcellation methods. SUVR values were extracted
without smoothing to preserve the highest possible resolution of the
PET data.
To obtain mean SUVR values for 11C-PiB PET, a standardized set
of ROIs obtained from FreeSurfer was dened on each hemisphere,
including the precuneus, frontal, orbitofrontal, parietal, temporal,
anterior cingulate, posterior cingulate, middle temporal cortices, and
the global cerebral cortex. A pre-established threshold of
11
C-PiB
SUVR >1.50 was applied to determine amyloid-β positivity, as
established in previous studies (Rowe etal., 2010). is threshold
value served as an indicator of elevated amyloid-β deposition in
the brain.
2.8 Statistical analysis
Statistical analyses were performed using SPSS (Version 28).
Stepwise linear regressions were used to examine the contributions of
regional tau burden to individual dierences in CDR-SB, MMSE, and
composite scores of delayed episodic memory and language. Separate

Rani et al. 10.3389/fnagi.2023.1272946
Frontiers in Aging Neuroscience 05 frontiersin.org
regressions were constructed for each dependent variable of interest.
Independent variables included age, sex, education, APOE4 status,
global cerebral cortex amyloid SUVR, FreeSurfer-derived EC tau
SUVR, as well as ASHS-derived ERC, BA35, and BA36 tau
SUVR. Right and le hemisphere ROIs were examined in separate
regressions to account for potential hemispheric dierences. is
approach was motivated by the statistically signicant variations
observed between the le and right FreeSurfer-derived EC tau SUVR
values [t (92) = −3.22, p < 0.002]. Stepwise regression utilizes a
mathematically driven approach to variable entry, whereby an
algorithm determines which set of variables maximizes the overall
proportion of explained variance. Independent variables were entered
into the model one at a time and subsequently removed if they did not
statistically improve the overall model. is allowed us to examine
which combination of measures predicted the highest proportion of
explained variance in each of the dependent variables.
Based on the results of the stepwise regressions, weperformed
secondary analyses using hierarchical linear regression. Unlike
stepwise models, hierarchical regression allows for user-determined
order of variable entry. For each dependent variable of interest,
weconstructed a hierarchical regression model utilizing the signicant
demographic predictors that emerged from the stepwise model, and
FIGURE1
Medial temporal lobe (MTL) subregion segmentation and tau PET image registration. (A) FreeSurfer parcellations of the entorhinal cortex, resampled in
MNI space (inset), with enlarged coronal view image highlighting the left EC (in yellow; volume 2079 mm3) and right EC (in red; volume 1859 mm3).
(B) MNI template images (inset) displaying the segmentation of the MTL subfields using ASHS, with enlarged coronal view image showing the left/right
ERC (in green; volume 566.8 mm3/volume 609.5 mm3), left/right Brodmann area 35 (in light blue; volume 623.8 mm3/volume 688 mm3), and left/right
Brodmann area 36 (in blue; volume 2,443 mm3/volume 2,466 mm3). (C) 18F-MK6240 PET image (inset), registered to the MNI template with a magnified
coronal view image highlighting registration between the images showing tau deposition in the ERC, BA35, and BA36in a representative CN subject.

Rani et al. 10.3389/fnagi.2023.1272946
Frontiers in Aging Neuroscience 06 frontiersin.org
then additionally entered the ASHS-derived ROIs, followed by the
FreeSurfer-derived EC ROI, allowing for a direct comparison of the
proportion of variance explained by the ASHS-derived versus
FreeSurfer-derived labels.
3 Results
e sample consisted of 82 CN individuals with a mean age of
68.33 ± 9.03 years (range: 49–88 years) and 11 individuals with MCI
with a mean age of 77.18 ± 6.15 years (range: 68–87 years). Sample
characteristics for all participants are summarized in Table 1. Sex
distribution was relatively balanced between the groups [𝜒
2
(1) = 0.004,
p = 0.95]. A signicant group dierence was observed for age [t
(91) = −3.15, p = 0.002], such that the MCI patients were, on average,
older than the CN participants. Group dierences were also observed
across all cognitive measures, with MCI patients exhibiting lower
performance compared to CN participants. e proportion of
individuals carrying the APOE4 allele was comparable between the two
diagnostic groups [𝜒
2
(1) = 0.06, p = 0.81], as was the distribution of
amyloid positivity across groups [𝜒
2
(1) = 0.88, p= 0.35]. Among the
CN individuals, 25 were classied as amyloid positive and 55 as
amyloid negative based on
11
C-PiB SUVR in the cerebral cortex.
Among the MCI individuals, 5 were amyloid positive and 6 were
amyloid negative. Group dierences in tau burden in specic MTL
subelds were only signicant in two subelds, le ASHS-derived ERC
[t (91) = −2.56, p = 0.01] and right ASHS-derived BA35 [t (91) = −2.73,
p = 0.008], with MCI patients exhibiting elevated SUVR values
compared to CN (Figure 2), although this analysis was likely
underpowered due to the small sample size in the MCI group.
Results from the stepwise linear regression analyses are presented
in Table2, and detailed results for each model are described below.
3.1 CDR Sum of Boxes
For CDR-SB, the right hemisphere stepwise linear regression
model identied right BA36 SUVR [ß = 0.68, t(86) = 4.21, p < 0.001;
Figure 3A], age [ß = 0.28, t (86) = 2.84, p = 0.006], and right BA35
SUVR [ß = −0.37, t(86) = −2.19, p = 0.03; Figure 3B] as signicant
predictors in the nal model. ese ndings indicate that higher tau
load in BA36 was associated with increased scores on the CDR-SB,
while lower levels of tau load in BA35 were associated with higher
CDR-SB scores. Notably, the addition of both of these variables
signicantly improved the overall proportion of explained variance.
Right BA36 tau load was the strongest single predictor [R
2
= 0.18,
F(1,88) = 19.62, p < 0.001], with subsequent steps demonstrating
signicant increases in the proportion of explained variance aer
adding age in Step2 [ΔR
2
= 0.05, ΔF(1,87) = 5.22, p = 0.03], and right
BA35in Step3 [ΔR2 = 0.04, ΔF (1,86) = 4.81, p = 0.03]. e strongest
model incorporated all three variables [R
2
= 0.27, F (3,86) = 10.58,
p < 0.001]. Importantly, the remaining variables, including FreeSurfer-
derived right EC SUVR, ASHS-derived right ERC SUVR, cerebral
cortex amyloid burden, APOE4 status, sex, and education did not
signicantly contribute to the model (p’s > 0.45). Furthermore, a
follow-up hierarchical linear regression analysis conrmed that the
inclusion of tau load from the FreeSurfer-derived right EC region did
not have a signicant impact on the proportion of explained variance
[ΔR
2
= 0.00, ΔF(1,88) = 0.002, p = 0.96; Figure 4A]. ese ndings
underscore the prominent role of right BA36 tau load, age, and right
BA35 tau load in predicting CDR-SB scores, while highlighting the
limited contribution of other variables considered in this model.
For the le hemisphere, tau load in le BA36 [ß = 0.40, t
(87) = 4.29, p < 0.001; Figure 3C] and age [ß = 0.25, t(87) = 2.71,
p = 0.008] were signicant predictors of CDR-SB. e positive
relationship between tau load in le BA36 and CDR-SB further
supports the notion that higher tau load in this region was associated
with higher CDR-SB scores. e addition of each variable signicantly
TABLE1 Participant characteristics.
Characteristic Cognitively
Normal
(n = 82)
MCI
(n = 11)
p value
Age, mean (SD) [range],
y
68.33 (9.03) [49–88] 77.18 (6.15)
[68–87]
0.002
Sex, No. (%)
Men 29 (35) 4 (36)
Wom en 53 (65) 7 (64)
Education, mean (SD), y 17.19 (2.18) 16.55 (2.98) 0.38
CDR-SB, mean (SD) 0.02 (0.12) 2.00 (1.34) < 0.001
MMSE score, mean (SD) 29.11 (0.93) 26.82 (2.18) 0.006
Logical memory
(delayed), mean (SD)
17.66 (3.68) 13.09 (4.72) < 0.001
CVLT long delayed Free
Recall, mean (SD)
14.13 (1.99) 9.73 (3.64) 0.002
Language composite
score (z-score), mean
(SD)
0.12 (0.37) −0.43 (0.40) < 0.001
Aβ status, No. (%) 0.35
Negative 55 (69) 6 (55)
Positive 25 (31) 5 (45)
APOE ε4 status, No. (%) 0.81
Non-carrier 49 (60) 7 (64)
Carrier 33 (40) 4 (36)
Le EC SUVR
(FreeSurfer) 1.92 (0.81) 2.79 (1.56) 0.10
Right EC SUVR
(FreeSurfer) 2.04 (0.90) 2.97 (1.73) 0.11
Le ERC SUVR (ASHS) 1.99 (0.95) 2.81 (1.39) 0.01
Right ERC SUVR
(ASHS) 2.07 (0.94) 2.88 (1.49) 0.11
Le BA35 SUVR (ASHS) 1.96 (0.92) 2.82 (1.73) 0.13
Right BA35 SUVR
(ASHS) 1.92 (0.81) 2.66 (1.09) 0.008
Le BA36 SUVR (ASHS) 1.75 (0.40) 2.45 (1.50) 0.15
Right BA36 SUVR
(ASHS)
1.78 (0.41) 2.48 (1.30)
0.11
Aβ, amyloid β; APOE, apolipoprotein E; BA, Brodmann area; CDR, Clinical Dementia
Rating; CVLT, California Verbal Learning Test; EC/ERC, entorhinal cortex; MCI, Mild
Cognitive Impairment; MMSE, Mini-Mental State Examination; SUVR, Standardized
Uptake Va lue Ratio.

Rani et al. 10.3389/fnagi.2023.1272946
Frontiers in Aging Neuroscience 07 frontiersin.org
FIGURE2
Patients with MCI show increased tau accumulation compared to control participants. Violin plots illustrating average distribution of 18F-MK6240 SUVR
in both FreeSurfer-derived EC and ASHS-derived MTL subregions (ERC, BA35, BA36) in cognitively normal (CN) older adults and MCI individuals in the
left hemisphere (A) and the right hemisphere (B). Patients with MCI showed significantly increased tau accumulation in the left ASHS-derived ERC and
right ASHS-derived BA35. The solid bars positioned in the center represent the median values, and the dotted bars indicate the interquartile range.
TABLE2 Stepwise regression models with 18F-MK6240 PET explaining variability in composite cognitive measures in right hemisphere models.
Dependent
variables
Independent
variables
βt-value FΔF R2ΔR2
CDR-SB
Step1 19.62*** 0.18
Right BA36 0.43 4.43***
Step2 12.89*** 5.22*0.23 0.05
Right BA36 0.39 4.10***
Age 0.22 2.29*
Step3 10.58*** 4.81*0.27 0.04
Right BA36 0.68 4.21***
Age 0.28 2.84**
Right BA35 −0.37 −2.19*
MMSE
Step1 9.19** 0.10
Right BA36 −0.31 −3.03**
Step2 7.17*** 4.76*0.14 0.05
Right BA36 −0.27 −2.70**
Age −0.22 −2.18*
Delayed memory composite score
Step1 4.02*4.02*0.04 0.04
Right BA36 −0.21 −2.01*
Language composite score
Step1 12.93*** 0.13
Education 0.36 3.60***
Step2 11.61*** 9.10** 0.21 0.08
Education 0.35 3.62***
Right BA36 −0.29 −3.02**
Variables entered into each model: age, sex, education, APOE4 carrier status, global amyloid burden, SUVR right EC (FreeSurfer), SUVR right ERC (ASHS), SUVR right BA36 (ASHS), SUVR
right BA35 (ASHS). *p< 0.05; **p< 0.01; ***p< 0.001.

Rani et al. 10.3389/fnagi.2023.1272946
Frontiers in Aging Neuroscience 08 frontiersin.org
improved the model, with Step1 indicating that le BA36 SUVR was
a signicant predictor [R
2
= 0.18, F(1,88) = 18.83, p < 0.001], followed
by a signicant increase in the proportion of explained variance aer
adding age in Step 2 [ΔR
2
= 0.06, ΔF(1,87) = 7.32, p = 0.008]. e
strongest model included both variables [R
2
= 0.24, F(2,87) = 13.75,
p < 0.001]. No other variables emerged as signicant predictors
FIGURE3
Relationships between tau load in medial temporal lobe (MTL) subregions and cognitive measures. Partial regression plots from stepwise linear
regression models with standardized residuals illustrating the association between tau load in MTL subregions and the variability observed in CDR-SB
(A–C), MMSE (D), and composite scores of delayed episodic memory (E) and language (F,G). The plots include shaded 95% confidence interval bands.
Separate models are shown for each dependent measure of cognition and for the left and right hemispheres. Models accounted for potential
contributions of age, sex, education, APOE4 status, and global amyloid burden.

Rani et al. 10.3389/fnagi.2023.1272946
Frontiers in Aging Neuroscience 09 frontiersin.org
(p’s > 0.08). Additionally, the follow-up hierarchical linear regression
conrmed that the inclusion of tau load in the le FreeSurfer-derived
EC did not signicantly change the proportion of explained variance
[ΔR
2
= 0.006, ΔF(1,88) = 0.67, p = 0.42; Figure4A]. Complete model
results are shown in Supplementary Table S1.
3.2 MMSE
With respect to MMSE, the nal stepwise model for the right
hemisphere MTL subregions revealed that tau load in right BA36
[ß = −0.27, t(87) = −2.70, p = 0.008; Figure3D] and age [ß = −0.22,
t(87) = −2.18, p = 0.03] were signicant predictors. ese ndings
suggest that higher tau load in right BA36 was signicantly associated
with poorer performance on the MMSE. Each variable signicantly
improved the model, with Step1 indicating that tau load in right BA36
was a signicant predictor [R
2
= 0.10, F(1,88) = 9.19, p = 0.003], and
with the inclusion of age in Step 2 signicantly increasing the
proportion of explained variance [ΔR
2
= 0.05, ΔF(1,87) = 4.76, p = 0.03].
e strongest model included both variables [R2 = 0.14, F(2,87) = 7.17,
p < 0.001]. Other considered variables, including sex, APOE4 status,
cerebral cortex amyloid burden, right FreeSurfer-derived EC SUVR,
right ASHS-derived ERC SUVR, and right BA35 SUVR, did not
signicantly contribute to the model (p’s > 0.06). Moreover, in the
FIGURE4
Tau burden in Brodmann area 36 is a significant predictor of CDR-SB and MMSE scores. Regression coecient betas (absolute values) from the
hierarchical regression analyses are plotted for variables of interest, with color coding based on FreeSurfer EC (red), ASHS-generated MTL subregions
(light blue and dark blue), age (orange), and education (green) for the CDR-SB (A), MMSE (B) and language composite scores (C). The error bars
represent 95% confidence intervals. Hierarchical models were constructed using significant demographic predictors from the previous stepwise
models, along with the larger inclusive FreeSurfer EC segmentation and the smaller MTL subregion segmentations provided through ASHS.
Importantly, tau PET burden in subregions, specifically BA36, consistently emerged as a significant predictor of CDR-SB and MMSE, while the larger
FreeSurfer EC label did not emerge as a significant predictor of cognition.

Rani et al. 10.3389/fnagi.2023.1272946
Frontiers in Aging Neuroscience 10 frontiersin.org
follow-up hierarchical regression, the inclusion of FreeSurfer-derived
right EC SUVR did not signicantly alter the proportion of explained
variance [ΔR
2
= 0.02, ΔF (1,88) = 1.72, p = 0.19; Figure4B]. By contrast,
the nal model for the le hemisphere MTL subregions revealed that
age was the only signicant predictor [ß = −0.26, t(88) = −2.57,
p = 0.01; see Supplementary Table S1], indicating an association
between age and individual variation in MMSE scores [R
2
= 0.07,
F(1,88) = 6.58, p = 0.01]. No signicant associations were observed
between MMSE scores and tau PET load in any of the le hemisphere
MTL subregions (p’s > 0.10).
3.3 Delayed episodic memory composite
score
For the delayed episodic memory composite score, the nal
stepwise model for the right hemisphere showed tau load in right
BA36 as the only signicant predictor [ß = −0.21, t (88) = −2.01,
p = 0.048; R
2
= 0.04, F (1,88) = 4.02, p = 0.048; Figure 3E]; the other
independent variables did not signicantly account for additional
variance (p’s > 0.51). is nding suggests that increased tau load in
right BA36 was associated with poorer delayed episodic memory.
Furthermore, in the follow-up hierarchical regression, the inclusion
of FreeSurfer-derived right EC SUVR did not signicantly change the
proportion of explained variance [ΔR
2
= 0.007, ΔF (1,89) = 0.65,
p = 0.42]. In the le hemisphere model, no variables emerged as
signicant predictors of the delayed episodic memory composite score.
3.4 Language composite score
For the language composite score, the nal stepwise model for the
right hemisphere MTL subelds included education [ß = 0.35,
t(87) = 3.62, p < 0.001] and right BA36 SUVR [ß = −0.29, t(87) = −3.02,
p = 0.003; Figure3F] as signicant predictors. e negative relationship
with BA36 indicates that higher tau load in this region corresponded
to lower language composite scores. e addition of both variables
signicantly improved the overall proportion of explained variance in
the model, with education emerging as a signicant predictor in Step1
[R
2
= 0.13, F(1,88) = 12.93, p < 0.001]. Subsequently, the addition of tau
load in right BA36in Step2 resulted in a signicant increase in the
proportion of explained variance [ΔR
2
= 0.08, ΔF (1,87) = 9.10,
p = 0.003]. e strongest model incorporated both [R
2
= 0.21,
F(2,87) = 11.61, p < 0.001], while the other considered variables did not
signicantly contribute to the model (p’s > 0.21). Furthermore, in the
follow-up hierarchical regression, the addition of FreeSurfer-derived
right EC SUVR did not signicantly alter the explained variance
[ΔR2 = 0.001, ΔF (1,88) = 0.07, p = 0.79; Figure4C].
Similarly, the nal stepwise model for the le hemisphere MTL
subelds showed education [ß = 0.36, t (87) = 3.78, p < 0.001] and tau
load in le BA36 were signicantly associated with language
composite scores [ß = −0.26, t (87) = −2.74, p = 0.008; Figure 3G].
Once again, the addition of each variable signicantly improved the
predictability of the overall model, with education included in Step1
[R
2
= 0.13, F (1,88) = 12.93, p < 0.001], and the addition of right BA36
tau load in Step2 [ΔR
2
= 0.07, ΔF (1,87) = 7.48, p = 0.008]. e strongest
model included both variables [R
2
= 0.20, F(2,87) = 10.68, p < 0.001],
and no other variables reached signicance for inclusion (p’s > 0.15).
e follow-up hierarchical regression analysis conrmed that the
addition of FreeSurfer-derived le EC tau load did not signicantly
enhance the proportion of explained variance in the nal model
[ΔR
2
= 0.02, ΔF(1,88) = 2.36, p = 0.13; Figure 4C]. ese ndings
emphasize the signicant role of education and tau load in BA36in
predicting the language composite score, while highlighting the
limited contribution of other regions and demographic variables.
Detailed results are shown in Supplementary Table S1.
3.5 Sensitivity analyses
To test whether our ndings were primarily driven by a subgroup
of participants, weperformed follow-up sensitivity analyses where
stepwise regression analyses were computed separately for each
diagnostic group. In the subsample of CN individuals only, right
BA35 SUVR [ß = 0.99, t (76) = 3.07, p = 0.003] and right ASHS-derived
ERC SUVR [ß = −0.80, t (76) = −2.48, p = 0.02] were signicant
predictors of CDR-SB. For the le hemisphere model, le FreeSurfer-
derived EC tau load signicantly contributed to the proportion of
explained variance in CDR-SB [ß = 0.24, t (77) = 2.19, p = 0.03]. No
signicant associations were found for MMSE models in the CN only
subsample. For both right and le hemisphere models, age was the
only signicant predictor to emerge from the delayed episodic
memory composite models [ß = 0.23, t (77) = 2.05, p = 0.04]. Similarly,
for both right and le hemisphere models, education was the only
signicant predictor of language composite scores [ß = 0.41, t
(77) = 3.97, p < 0.001]. Detailed results are shown in
Supplementary Table S2.
In the subsample of MCI patients only, no signicant predictors
were found in stepwise models for either hemisphere predicting
CDR-SB or delayed episodic memory composite. For both right and
le hemisphere models, sex signicantly contributed to the
proportion of explained variance in MMSE scores [ß = 0.75, t
(9) = 3.42, p < 0.008]. Finally, right BA36 SUVR was signicantly
associated with language composite scores in the MCI only subsample
[ß = −0.71, t(9) = −2.98, p = 0.02]. It is important to note the
exploratory nature of the MCI only subgroup analysis, as it is
underpowered due to the very small number of MCI patients in this
study. e overall pattern of results from these sensitivity analyses is
consistent with the overall ndings and further supports the utility of
MTL subregional analyses, as subtle dierences in patterns of tau
accumulation may emerge over time or across individuals. Complete
results are shown in Supplementary Table S3.
4 Discussion
is study evaluated contributions of regional tau burden in
specic MTL subregions, as measured by
18
F-MK6240 PET, to
individual variability in cognition in order to determine whether tau
burden localized to subregions of the entorhinal cortex provides
additional specicity compared to tau burden quantied in larger
segmentations. Taking advantage of the high binding specicity of this
second-generation tracer, in combination with the application of
advanced registration and segmentation methods provided by the
ANTS and ASHS soware packages, this approach enabled
quantication of in vivo tau burden within specic subregions of the

Rani et al. 10.3389/fnagi.2023.1272946
Frontiers in Aging Neuroscience 11 frontiersin.org
MTL. Results showed that tau burden, specically in le and right
BA36 and right BA35, as well as age, were each independently related
to individual dierences in CDR-SB scores. Similarly, tau burden in
right BA36 and age each uniquely contributed to the proportion of
explained variance in MMSE scores, while le and right BA36 tau
burden and education were associated with language composite
scores, and right BA36 tau burden signicantly predicted delayed
episodic memory composite scores.
While the stepwise regression analyses aimed to identify the best
overall combination of variables to explain individual dierences in
cognition, irrespective of the origin of the ROIs (i.e., FreeSurfer vs.
ASHS), wewere also interested in directly comparing the relative
contributions of the FreeSurfer EC ROI to the ASHS MTL ROIs. To
address this question, hierarchical regression analyses were used to
ensure that the FreeSurfer EC ROI was retained in the models
regardless of the predictor’s statistical signicance. Importantly, the
addition of the more inclusive ROI encompassing a broader
parcellation of the entorhinal cortex did not signicantly contribute
to the explained variance in any of the cognitive measures. Together,
these ndings suggest that quantication of tau burden within more
localized subregions of the MTL may better account for individual
dierences in cognition in non-demented older adults.
ese ndings are not only consistent with previous reports
showing signicantly increased accumulation of tau deposition in
MTL subregions using
18
F-AV1451 PET imaging (Johnson et al.,
2016), but further localize early deposition of tau to the right BA35
subregion. BA35 largely overlaps with what Braak and colleagues
referred to as TEC, where they observed some of the earliest
neurobrillary tangle accumulation in patients with MCI due to AD
(Braak and Braak, 1990; Braak etal., 2006; Kaufman etal., 2018). e
results further demonstrate signicant relationships between tau
burden in MTL subregions and cognition in non-demented older
adults. Specically, tau burden in BA36 was associated with individual
dierences across all cognitive assessments, indicating the importance
of this region in the interaction between pathological tau accumulation
and cognition.
Additionally, there was a signicant association between BA35
and CDR-SB scores (a measure designed to reect both cognition and
function). It is important to note that the direction of this relationship
showed that lower levels of tau load in BA35 were associated with
higher CDR-SB scores. Given the strong positive association between
tau load in BA36 and CDR-SB, this negative relationship with tau load
in BA35 likely reects the topographical progression of tau
accumulation with disease progression described in Braak staging.
e observed relationship between tau burden and memory aligns
with the functional role of the brain regions where this eect was
observed. e temporal lobe plays a critical role in episodic memory
function but also plays an important role in language function, so it is
possible that the language nding in our data may reect some
interplay between dierent temporal lobe regions and cognitive
functions. Although few studies have examined subregion-specic tau
pathology accumulation using PET imaging (Adams et al., 2019;
Berron etal., 2021; Chen etal., 2021; Ge et al., 2021), functional
specialization of MTL subregions has been previously reported using
functional MRI (Maass etal., 2015; Adams etal., 2021). Studies of CN
adults have shown that the lateral EC and perirhinal cortex support
encoding of object and content information, while medial EC and
parahippocampal regions are thought to facilitate encoding of context
and spatial information (Knierim etal., 2014; Reagh and Yassa, 2014;
Yeung etal., 2017; Berron etal., 2018; Montchal etal., 2019).
Similarly, studies of patients with MCI have shown reduced
volume in the lateral EC, as well as reduced activation in the lateral
EC, when compared to CN participants (Reagh etal., 2018; Tra n
etal., 2021, 2022). ese changes in the lateral EC are associated
with the memory performance of individuals with MCI, which are
not observed in association with the medial EC. ese ndings are
consistent with the results reported here, as lateral EC overlaps
primarily with BA35 and partially with BA36, where associations
with tau and cognition were observed, while the medial EC
corresponds to the ERC label, where such associations were not
present. Given the observed functional specialization and selective
engagement of EC subregions in patients with MCI, the
examination of tau burden in more localized subregions of the
MTL may provide new opportunities to understand individual
dierences and improve the identication of non-demented older
adults who are on a trajectory of decline due to AD. However, it is
important to note that this study measured tau SUVR in MTL
regions, rather than functional activation, cortical thickness, or
volume; as such, it is not necessarily expected to see cognitive
domain-specic eects associated with tau burden. Given the
established association between the progression of tau accumulation
and overall decline in cognition throughout disease progression, it
seems reasonable that individual dierences in tau burden in these
early Braak regions would beassociated with individual dierences
in overall cognition.
Several limitations of the current study should be noted.
First, the ROIs used in the current study compared the FreeSurfer
and ASHS softwares, which each employ somewhat different
boundaries for the EC with none of the segmented regions in
either software being fully consistent with the TEC used in the
post-mortem studies showing the earliest stages of tau. Functional
studies of the EC have employed a lateral versus medial
distinction; however, those boundaries are similarly not
consistent across studies. A consensus of landmarks and
terminology based on both anatomical and functional studies is
needed to facilitate comparisons across studies and further assess
the functional correlates of tau accumulation in subregions of the
EC. Second, the present analyses are cross-sectional, based on a
single assessment of tau load, limiting claims with respect to
changes in tau spread and cognition over time. However, data
collection in this cohort is ongoing, making longitudinal analyses
possible in future studies. Finally, it is important to note that the
study only included a small number of MCI patients, limiting our
ability to detect potential group differences or make claims
regarding the MCI subgroup. To determine whether the pattern
of results was driven by this small group of MCI patients, a series
of follow-up sensitivity analyses was performed limiting our
stepwise regressions to include only CN individuals or only MCI
individuals. Importantly, the overall pattern of results within the
CN subgroup was consistent with the primary findings, indicating
that the relationships between tau burden in MTL subregions and
cognition were not solely driven by the MCI subgroup of patients.
In these constrained analyses, the localized subregions remained
informative, supporting the conclusion that consideration of
specific subregions may enhance sensitivity for the identification
of individuals on the AD-trajectory.

Rani et al. 10.3389/fnagi.2023.1272946
Frontiers in Aging Neuroscience 12 frontiersin.org
5 Conclusion
e current study examined associations between regional tau
burden in specic MTL subregions and relationships with individual
dierences in cognition in non-demented older adults. e ndings
indicate that applying advanced registration and segmentation methods
to tau PET images can achieve enhanced visualization and quantication
of tau uptake localized to subregions within the MTL. Estimates of tau
load in specic MTL subregions furthermore explain variance in
individual dierences in CDR-SB scores and across multiple cognitive
domains. Together, these ndings suggest that tau accumulation
localized to subregions of the entorhinal cortex provides additional
specicity compared to accumulation quantied in larger segmentations,
therefore highlighting the importance of examining associations between
localized tau burden and cognitive variability for improving the
characterization of individual disease trajectories and identifying those
individuals who are on a trajectory of decline due to AD.
Data availability statement
e raw data supporting the conclusions of this article will
bemade available by the authors, without undue reservation.
Ethics statement
e studies involving humans were approved by Johns Hopkins
University Institutional Review Board. e studies were conducted in
accordance with the local legislation and institutional requirements. e
participants provided their written informed consent to participate in
this study.
Author contributions
NR: Data curation, Formal analysis, Methodology, Visualization,
Writing – original dra, Writing – review & editing. KA: Formal analysis,
Methodology, Visualization, Writing – review & editing, Data curation,
Writing – original dra. CC-L: Methodology, Writing – review & editing.
CS: Writing – review & editing, Data curation, Validation. AS:
Conceptualization, Data curation, Methodology, Project administration,
Writing – review & editing. CP: Conceptualization, Data curation,
Methodology, Project administration, Writing – review & editing. YZ:
Methodology, Writing – review & editing. MA: Conceptualization, Data
curation, Funding acquisition, Investigation, Methodology, Project
administration, Resources, Supervision, Validation, Writing – review &
editing. AB: Conceptualization, Data curation, Funding acquisition,
Investigation, Methodology, Project administration, Resources,
Supervision, Validation, Writing – review & editing.
Funding
e author(s) declare nancial support was received for the
research, authorship, and/or publication of this article. is work was
supported by grants from the National Institutes of Health (U19-
AG033655 and P41-EB031771).
Acknowledgments
e authors thank the entire BIOCARD study team at Johns
Hopkins University for their support, the BIOCARD participants for
continuing to participate in the study, and the Geriatric Psychiatry
Branch of the intramural program of the NIMH who initiated this
study (PI: Trey Sunderland).
Conflict of interest
e authors declare that the research was conducted in the
absence of any commercial or nancial relationships that could
beconstrued as a potential conict of interest.
Publisher’s note
All claims expressed in this article are solely those of the authors
and do not necessarily represent those of their aliated organizations,
or those of the publisher, the editors and the reviewers. Any product
that may be evaluated in this article, or claim that may be made by its
manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
e Supplementary material for this article can befound online
at: https://www.frontiersin.org/articles/10.3389/fnagi.2023.1272946/
full#supplementary-material
References
Adams, J. N., Maass, A., Berron, D., Harrison, T. M., Baker, S. L., omas, W. P., et al.
(2021). Reduced repetition suppression in aging is driven by tau-related hyperactivity in
medial temporal lobe. J. Neurosci. 41, 3917–3931. doi: 10.1523/JNEUROSCI.2504-20.2021
Adams, J. N., Maass, A., Harrison, T. M., Baker, S. L., and Jagust, W. J. (2019). Cortical
tau deposition follows patterns of entorhinal functional connectivity in aging. elife
8:e49132. doi: 10.7554/eLife.49132
Albert, M. S., DeKosky, S. T., Dickson, D., Dubois, B., Feldman, H. H., Fox, N. C., et al.
(2011). e diagnosis of mild cognitive impairment due to Alzheimer’s disease:
recommendations from the National Institute on Aging-Alzheimer’s association
workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 7,
270–279. doi: 10.1016/j.jalz.2011.03.008
Albert, M., Soldan, A., Gottesman, R., McKhann, G., Sacktor, N., Farrington, L., et al.
(2014). Cognitive changes preceding clinical symptom onset of mild cognitive
impairment and relationship to ApoE genotype. Curr. Alzheimer Res. 11, 773–784. doi:
10.2174/156720501108140910121920
Alm, K. H., Soldan, A., Pettigrew, C., Faria, A. V., Hou, X., Lu, H., et al. (2022).
Structural and functional brain connectivity uniquely contribute to episodic memory
performance in older adults. Front. Aging Neurosci. 14:951076. doi: 10.3389/
fnagi.2022.951076
Avants, B. B., Epstein, C. L., Grossman, M., and Gee, J. C. (2008). Symmetric
dieomorphic image registration with cross-correlation: evaluating automated labeling
of elderly and neurodegenerative brain. Med. Image Anal. 12, 26–41. doi: 10.1016/j.
media.2007.06.004
Avants, B. B., Tustison, N. J., Stauer, M., S ong, G., Wu, B., and Gee, J. C. (2014). e
insight ToolKit image registration framework. Front. Neuroinform. 8:44. doi: 10.3389/
fninf.2014.00044

Rani et al. 10.3389/fnagi.2023.1272946
Frontiers in Aging Neuroscience 13 frontiersin.org
Bao, W., Jia, H., Finnema, S., Cai, Z., Carson, R. E., and Huang, Y. H. (2017). PET
imaging for early detection of Alzheimer’s Disease: from pathologic to physiologic
biomarkers. PET Clinics 12, 329–350. doi: 10.1016/j.cpet.2017.03.001
Berron, D., Neumann, K., Maass, A., Schütze, H., Fliessbach, K., Kiven, V., et al.
(2018). Age-related functional changes in domain-specic medial temporal lobe
pathways. Neurobiol. Aging 65, 86–97. doi: 10.1016/j.neurobiolaging.2017.12.030
Berron, D., Vogel, J. W., Insel, P. S., Pereira, J. B., Xie, L., Wisse, L. E. M., et al. (2021).
Early stages of tau pathology and its associations with functional connectivity, atrophy
and memory. Brain 144, 2771–2783. doi: 10.1093/brain/awab114
Betthauser, T. J., Koscik, R. L., Jonaitis, E. M., Allison, S. L., Cody, K. A.,
Erickson, C. M., et al. (2020). Amyloid and tau imaging biomarkers explain cognitive
decline from late middle-age. Brain 143, 320–335. doi: 10.1093/brain/awz378
Blennow, K., Hampel, H., Weiner, M., and Zetterberg, H. (2010). Cerebrospinal uid
and plasma biomarkers in Alzheimer disease. Nat. Rev. Neurol. 6, 131–144. doi: 10.1038/
nrneurol.2010.4
Braak, H., Alafuzo, I., Arzberger, T., Kretzschmar, H., and Del Tredici, K. (2006).
Staging of Alzheimer disease-associated neurobrillary pathology using paran
sections and immunocytochemistry. Acta Neuropathol. 112, 389–404. doi: 10.1007/
s00401-006-0127-z
Braak, H., and Braak, E. (1990). Neurobrillary changes conned to the entorhinal
region and an abundance of cortical amyloid in cases of presenile and senile dementia.
Acta Neuropathol. 80, 479–486. doi: 10.1007/BF00294607
Braak, H., and Braak, E. (1991). Neuropathological stageing of Alzheimer-related
changes. Acta Neuropathol. 82, 239–259. doi: 10.1007/BF00308809
Brier, M. R., Gordon, B., Friedrichsen, K., McCarthy, J., Stern, A., Christensen, J., et al.
(2016). Tau and Aβ imaging, CSF measures, and cognition in Alzheimer’s disease. Sci.
Transl. Med. 8:338ra66-338ra66. doi: 10.1126/scitranslmed.aaf2362
Chen, X., Cassady, K. E., Adams, J. N., Harrison, T. M., Baker, S. L., and Jagust, W. J.
(2021). Regional tau eects on prospective cognitive change in cognitively Normal older
adults. J. Neurosci. O. J. Soc. Neurosci. 41, 366–375. doi: 10.1523/
JNEUROSCI.2111-20.2020
Chiotis, K., Savitcheva, I., Poulakis, K., Saint-Aubert, L., Wall, A., Antoni, G., et al.
(2021). [18F] THK5317 imaging as a tool for predicting prospective cognitive decline
in Alzheimer’s disease. Mol. Psychiatry 26, 5875–5887. doi: 10.1038/s41380-020-0815-4
Dale, A. M., Fischl, B., and Sereno, M. I. (1999). Cortical surface-based analysis. I.
Segmentation and surface reconstruction. Neuro Image 9, 179–194. doi: 10.1006/
nimg.1998.0395
Dickerson, B. C., and Eichenbaum, H. (2010). e episodic memory system: Neurocircuitry
and disorders. Neuropsychopharmacology 35, 86–104. doi: 10.1038/npp.2009.126
Fischl, B. (2012). Free Surfer. NeuroImage 62, 774–781. doi: 10.1016/j.
neuroimage.2012.01.021
Fischl, B., Salat, D. H., Busa, E., Albert, M., Dieterich, M., Haselgrove, C., et al. (2002).
Whole brain segmentation: automated labeling of neuroanatomical structures in the
human brain. Neuron 33, 341–355. doi: 10.1016/s0896-6273(02)00569-x
Fischl, B., Sereno, M. I., Tootell, R. B. H., and Dale, A. M. (1999). High-resolution
intersubject averaging and a coordinate system for the cortical surface. Hum. Brain
Mapp. 8, 272–284. doi: 10.1002/(SICI)1097-0193(1999)8:4<272::AID-HBM10>3.0.CO;2-4
Fu, J. F., Lois, C., Sanchez, J., Becker, J. A., Rubinstein, Z. B., ibault, E., et al. (2023).
Kinetic evaluation and assessment of longitudinal changes in reference region and
extracerebral [18F]MK-6240 PET uptake. J. Cereb. Blood Flow Metab. 43, 581–594. doi:
10.1177/0271678X221142139
Ge, X., Zhang, D., Qiao, Y., Zhang, J., Xu, J., and Zheng, Y. (2021). Association of tau
Pathology with Clinical Symptoms in the subelds of hippocampal formation. Front.
Aging Neurosci. 13:672077. doi: 10.3389/fnagi.2021.672077
Gogola, A., Minhas, D. S., Villemagne, V. L., Cohen, A. D., Mountz, J. M.,
Pascoal, T. A., et al. (2022). Direct comparison of the tau PET tracers 18F-Flortaucipir
and 18F-MK-6240 in human subjects. J. Nucl. Med. 63, 108–116. doi: 10.2967/
jnumed.120.254961
Greve, D. N., and Fischl, B. (2009). Accurate and robust brain image alignment using
boundary-based registration. NeuroImage 48, 63–72. doi: 10.1016/j.
neuroimage.2009.06.060
Hall, B., Mak, E., Cervenka, S., Aigbirhio, F. I., Rowe, J. B., and O’Brien, J. T. (2017).
In vivo tau PET imaging in dementia: pathophysiology, radiotracer quantication, and
a systematic review of clinical ndings. Ageing Res. Rev. 36, 50–63. doi: 10.1016/j.
arr.2017.03.002
Hostetler, E. D., Walji, A. M., Zeng, Z., Miller, P., Bennacef, I., Salinas, C., et al. (2016).
Preclinical characterization of 18F-MK-6240, a promising PET tracer for in vivo
quantication of human neurobrillary tangles. J Nuclear Med: Ocial Publication, Soc
Nuclear Med 57, 1599–1606. doi: 10.2967/jnumed.115.171678
Hudson, H. M., and Larkin, R. S. (1994). Accelerated image reconstruction using
ordered subsets of projection data. IEEE Trans. Med. Imaging 13, 601–609. doi:
10.1109/42.363108
Jenkinson, M., Bannister, P., Brady, M., and Smith, S. (2002). Improved optimization
for the robust and accurate linear registration and motion correction of brain images.
NeuroImage 17, 825–841. doi: 10.1016/s1053-8119(02)91132-8
Jenkinson, M., and Smith, S. (2001). A global optimisation method for robust ane
registration of brain images. Med. Image Anal. 5, 143–156. doi: 10.1016/
s1361-8415(01)00036-6
Johnson, K. A., Schultz, A., Betensky, R. A., Becker, J. A., Sepulcre, J., Rentz, D., et al.
(2016). Tau positron emission tomographic imaging in aging and early Alzheimer
disease. Ann. Neurol. 79, 110–119. doi: 10.1002/ana.24546
Kaufman, S. K., Del Tredici, K., omas, T. L., Braak, H., and Diamond, M. I. (2018).
Tau seeding activity begins in the transentorhinal/entorhinal regions and anticipates
phospho-tau pathology in Alzheimer’s diseas e and PART. Acta Neuropathol. 136, 57–67.
doi: 10.1007/s00401-018-1855-6
Klunk, W. E., Engler, H., Nordberg, A., Wang, Y., Blomqvist, G., Holt, D. P., et al.
(2004). Imaging brain amyloid in Alzheimer’s disease with Pittsburgh compound-B.
Ann. Neurol. 55, 306–319. doi: 10.1002/ana.20009
Knierim, J. J., Neunuebel, J. P., and Deshmukh, S. S. (2014). Functional correlates of
the lateral and medial entorhinal cortex: objects, path integration and local–global
reference frames. Philosophical Transactions of the R Soc B: Biological Sci 369:20130369.
doi: 10.1098/rstb.2013.0369
Krishnadas, N., Doré, V., Robertson, J. S., Ward, L., Fowler, C., Masters, C. L., et al.
(2023). Rates of regional tau accumulation in ageing and across the Alzheimer’s disease
continuum: an AIBL 18F-MK6240 PET study. EBioMedicine 88:104450. doi: 10.1016/j.
ebiom.2023.104450
Liu, E., Schmidt, M. E., Margolin, R., Sperling, R., Koeppe, R., Mason, N. S., et al.
(2015). Amyloid-β 11 C-PiB-PET imaging results from 2 randomized bapineuzumab
phase 3 AD trials. Neurology 85, 692–700. doi: 10.1212/WNL.0000000000001877
Lohith, T. G., Bennacef, I., Vandenberghe, R., Vandenbulcke, M., Salinas, C. A.,
Declercq, R., et al. (2019). Brain imaging of Alzheimer dementia patients and elderly
controls with 18F-MK-6240, a PET tracer targeting neurobrillary tangles. J. Nucl. Med.
60, 107–114. doi: 10.2967/jnumed.118.208215
Maass, A., Berron, D., Libby, L. A., Ranganath, C., and Düzel, E. (2015). Functional
subregions of the human entorhinal cortex. elife 4, 1–20. doi: 10.7554/eLife.06426
Maass, A., Lockhart, S. N., Harrison, T. M., Bell, R. K., Mellinger, T., Swinnerton, K.,
et al. (2018). Entorhinal tau pathology, episodic memory decline, and neurodegeneration
in aging. J. Neurosci. 38, 530–543. doi: 10.1523/JNEUROSCI.2028-17.2017
Malarte, M.-L., Nordberg, A., and Lemoine, L. (2021). Characterization of MK6240,
a tau PET tracer, in autopsy brain tissue from Alzheimer’s disease cases. Eur. J. Nucl.
Med. Mol. Imaging 48, 1093–1102. doi: 10.1007/s00259-020-05035-y
Manjon, J., Coupé, P., Buades, A., Fonov, V., Collins, L., and Robles, M. (2010). Non-
local MRI upsampling. Med. Image Anal. 14, 784–792. doi: 10.1016/j.media.2010.05.010
McKhann, G. M., Knopman, D. S., Chertkow, H., Hyman, B. T., Jack, C. R.,
Kawas, C. H., et al. (2011). e diagnosis of dementia due to Alzheimer’s disease:
recommendations from the National Institute on Aging-Alzheimer’s association
workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 7,
263–269. doi: 10.1016/j.jalz.2011.03.005
Montchal, M. E., Reagh, Z. M., and Yassa, M. A. (2019). Precise temporal memories
are supported by the lateral entorhinal cortex in humans. Nat. Neurosci. 22, 284–288.
doi: 10.1038/s41593-018-0303-1
Morris, J. C. (1993). e Clinical dementia rating (CDR): current version and scoring
rules. Neurology 43, 2412–2414. doi: 10.1212/wnl.43.11.2412-a
Nelson, P. T., Alafuzo, I., Bigio, E. H., Bouras, C., Braak, H., Cairns, N. J., et al. (2012).
Correlation of Alzheimer Disease Neuropathologic changes with cognitive status: a
review of the literature. J. Neuropathol. Exp. Neurol. 71, 362–381. doi: 10.1097/
NEN.0b013e31825018f7
Ossenkoppele, R., Lyoo, C. H., Jester-Broms, J., Sudre, C. H., Cho, H., Ryu, Y. H., et al.
(2020). Assessment of demographic, genetic, and imaging variables associated with
brain resilience and cognitive resilience to pathological tau in patients with Alzheimer
disease. JAMA Neurol. 77, 632–642. doi: 10.1001/jamaneurol.2019.5154
Ossenkoppele, R., Schonhaut, D. R., Schöll, M., Lockhart, S. N., Ayakta, N.,
Baker, S. L., et al. (2016). Tau PET patterns mirror clinical and neuroanatomical
variability in Alzheimer’s disease. Brain 139, 1551–1567. doi: 10.1093/brain/aww027
Ossenkoppele, R., Smith, R., Mattsson-Carlgren, N., Groot, C., Leuzy, A.,
Strandberg, O., et al. (2021). Accuracy of tau positron emission tomography as a
prognostic marker in preclinical and prodromal Alzheimer disease: a head-to-head
comparison against amyloid positron emission tomography and magnetic resonance
imaging. JAMA Neurol. 78, 961–971. doi: 10.1001/jamaneurol.2021.1858
Pascoal, T. A., Shin, M., Kang, M. S., Chamoun, M., Chartrand, D., Mathotaarachchi, S.,
et al. (2018). In vivo quantication of neurobrillary tangles with [18F]MK-6240.
Alzheimers Res. er. 10:74. doi: 10.1186/s13195-018-0402-y
Pletnikova, O., Kageyama, Y., Rudow, G., LaClair, K. D., Albert, M., Crain, B. J., et al.
(2018). e spectrum of preclinical Alzheimer’s disease pathology and its modulation
by ApoE genotype. Neurobiol. Aging 71, 72–80. doi: 10.1016/j.neurobiolaging.2018.07.007
Price, J. C., Klunk, W. E., Lopresti, B. J., Lu, X., Hoge, J. A., Ziolko, S. K., et al.
(2005). Kinetic modeling of amyloid binding in humans using PET imaging and
Pittsburgh compound-B. J. Cereb. Blood Flow Metab. 25, 1528–1547. doi: 10.1038/
sj.jcbfm.9600146
Reagh, Z. M., Noche, J. A., Tustison, N. J., Delisle, D., Murray, E. A., and Yassa, M. A.
(2018). Functional imbalance of anterolateral entorhinal cortex and hippocampal

Rani et al. 10.3389/fnagi.2023.1272946
Frontiers in Aging Neuroscience 14 frontiersin.org
dentate/CA3 underlies age-related object pattern separation decits. Neuron 97,
1187–1198.e4. doi: 10.1016/j.neuron.2018.01.039
Reagh, Z. M., and Yassa, M. A. (2014). Object and spatial mnemonic interference
dierentially engage lateral and medial entorhinal cortex in humans. Proc. Natl. Acad.
Sci. U. S. A. 111, E4264–E4273. doi: 10.1073/pnas.1411250111
Ricci, M., Cimini, A., Camedda, R., Chiaravalloti, A., and Schillaci, O. (2021). Tau
biomarkers in dementia: positron emission tomography radiopharmaceuticals in
Tauopathy assessment and future perspective. Int. J. Mol. Sci. 22:13002. doi: 10.3390/
ijms222313002
Rousset, O. G., Ma, Y., and Evans, A. C. (1998). Correction for partial volume eects
in PET: principle and validation. J Nuclear Med: O Publ, Soc Nuclear Med. 39, 904–911.
Rowe, C. C., Ellis, K. A., Rimajova, M., Bourgeat, P., Pike, K. E., Jones, G., et al. (2010).
Amyloid imaging results from the Australian imaging, biomarkers and lifestyle (AIBL)
study of aging. Neurobiol. Aging 31, 1275–1283. doi: 10.1016/j.neurobiolaging.2010.04.007
Scheinin, N. M., Aalto, S., Koikkalainen, J., Lötjönen, J., Karrasch, M., Kemppainen, N.,
et al. (2009). Follow-up of [11C] PIB uptake and brain volume in patients with alzheimer
disease and controls. Neurology 73, 1186–1192. Scopus. doi: 10.1212/
WNL.0b013e3181bacf1b
Ségonne, F., Pacheco, J., and Fischl, B. (2007). Geometrically accurate topology-
correction of cortical surfaces using nonseparating loops. IEEE Trans. Med. Imaging 26,
518–529. doi: 10.1109/TMI.2006.887364
Shidahara, M., omas, B. A., Okamura, N., Ibaraki, M., Matsubara, K., Oyama, S.,
et al. (2017). A comparison of ve partial volume correction methods for tau and
amyloid PET imaging with [18F] THK5351 and [11C]PIB. Ann. Nucl. Med. 31, 563–569.
doi: 10.1007/s12149-017-1185-0
Soldan, A., Pettigrew, C., Fagan, A. M., Schindler, S. E., Moghekar, A., Fowler, C., et al.
(2019). ATN proles among cognitively normal individuals and longitudinal cognitive
outcomes. Neurology 92, e1567–e1579. doi: 10.1212/WNL.0000000000007248
Sperling, R. A., Aisen, P. S., B eckett, L. A., Bennett, D. A., Cra, S., Fagan, A. M., et al.
(2011). Toward dening the preclinical stages of Alzheimer’s disease: recommendations
from the National Institute on Aging-Alzheimer’s associ ation workgroups on diagnostic
guidelines for Alzheimer’s disease. Alzheimers Dement. 7, 280–292. doi: 10.1016/j.
jalz.2011.03.003
Taylor, K. I., and Probst, A. (2008). Anatomic localization of the transentorhinal
region of the perirhinal cortex. Neurobiol. Aging 29, 1591–1596. doi: 10.1016/j.
neurobiolaging.2007.03.024
omas, B. A., Erlandsson, K., Modat, M., urell, L., Vandenberghe, R., Ourselin, S.,
et al. (2011). e importance of appropriate partial volume correction for PET
quantication in Alzheimer’s disease. Eur. J. Nucl. Med. Mol. Imaging 38, 1104–1119. doi:
10.1007/s00259-011-1745-9
Tran, T. T., Speck, C. L., Gallagher, M., and Bakker, A. (2022). Lateral entorhinal cortex
dysfunction in amnestic mild cognitive impairment. Neurobiol. Aging 112, 151–160. doi:
10.1016/j.neurobiolaging.2021.12.008
Tran, T., Tobin, K. E., Block, S. H., Puliyadi, V., Gallagher, M., and Bakker, A. (2021).
Eect of aging diers for memory of object identity and object position within a spatial
context. Learn. Mem. 28, 239–247. doi: 10.1101/lm.053181.120
Tustison, N. J., Cook, P. A., Klein, A., Song, G., Das, S. R., Duda, J. T., et al. (2014).
Large-scale evaluation of ANTs and FreeSurfer cortical thickness measurements.
NeuroImage 99, 166–179. doi: 10.1016/j.neuroimage.2014.05.044
Wang, H., Das, S. R., Suh, J. W., Altinay, M., Pluta, J., Craige, C., et al. (2011). A
learning-based wrapper method to correct systematic errors in automatic image
segmentation: consistently improved performance in hippocampus, cortex and brain
segmentation. NeuroImage 55, 968–985. doi: 10.1016/j.neuroimage.2011.01.006
Wang, H., Suh, J. W., Das, S. R., Pluta, J. B., Craige, C., and Yushkevich, P. A. (2013).
Multi-atlas segmentation with joint label fusion. IEEE Trans. Pattern Anal. Mach. Intell.
35, 611–623. doi: 10.1109/TPAMI.2012.143
Xie, L., Wisse, L. E. M., Pluta, J., de Flores, R., Piskin, V., Manjón, J. V., et al. (2019).
Automated segmentation of medial temporal lobe subregions on invivo T1-weighted
MRI in early stages of Alzheimer’s disease. Hum. Brain Mapp. 40, 3431–3451. doi:
10.1002/hbm.24607
Yeung, L.-K., Olsen, R. K., Bild-Enkin, H. E. P., D’Angelo, M. C., Kacollja, A.,
McQuiggan, D. A., et al. (2017). Anterolateral entorhinal cortex volume predicted by
altered intra-item Congural processing. J. Neurosci. O. J. Soc. Neurosci. 37, 5527–5538.
doi: 10.1523/JNEUROSCI.3664-16.2017
Yushkevich, P. A., Gao, Y., and Gerig, G. (2016). ITK-SNAP: an interactive tool for
semi-automatic segmentation of multi-modality biomedical images. 2016 38th annual
international conference of the IEEE engineering in medicine and biology society
(EMBC), 3342–3345.
Yushkevich, P. A., Pluta, J. B., Wang, H., Xie, L., Ding, S., Gertje, E. C., et al. (2014).
Automated volumetry and regional thickness analysis of hippocampal subelds and
medial temporal cortical structures in mild cognitive impairment. Hum. Brain Mapp.
36, 258–287. doi: 10.1002/hbm.22627