Data Aggrega)on, Synthesis and Replica)on: Why Res)ng State fMRI Is and Is Not Ideal Michael P. Milham, MD, PhD What Makes R-­‐fMRI the “Low-­‐Hanging Fruit” for Data AggregaBon? •  Bypasses potenBal sources of variaBon associated with task probes •  Commonly included as an add-­‐on in task acBvaBon studies –  decreases perceived value –  Increases willingness to share •  Striking similarity in networks observed across laboratories What Makes R-­‐fMRI Less Than Ideal for AggregaBon Yan et al., 2013b
Lessons Learned From the FCP/INDI Experience !
http://fcon_1000.projects.nitrc.org
#1 Aggregate R-­‐fMRI Data Analysis is Possible Toward discovery science of human brain function
Bharat B. Biswala, Maarten Mennesb, Xi-Nian Zuob, Suril Gohela, Clare Kellyb, Steve M. Smithc, Christian F. Beckmannc,
Jonathan S. Adelsteinb, Randy L. Bucknerd, Stan Colcombee, Anne-Marie Dogonowskif, Monique Ernstg, Damien Fairh,
Michelle Hampsoni, Matthew J. Hoptmanj, James S. Hydek, Vesa J. Kiviniemil, Rolf Kötterm, Shi-Jiang Lin, Ching-Po Lino,
Mark J. Lowep, Clare Mackayc, David J. Maddenq, Kristoffer H. Madsenf, Daniel S. Marguliesr, Helen S. Maybergs,
Katie McMahont, Christopher S. Monku, Stewart H. Mostofskyv, Bonnie J. Nagelw, James J. Pekarx, Scott J. Peltiery,
Steven E. Petersenz, Valentin Riedlaa, Serge A. R. B. Romboutsbb, Bart Rypmacc, Bradley L. Schlaggardd, Sein Schmidtee,
Rachael D. Seidlerff,u, Greg J. Sieglegg, Christian Sorghh, Gao-Jun Tengii, Juha Veijolajj, Arno Villringeree,kk,
Martin Walterll, Lihong Wangq, Xu-Chu Wengmm, Susan Whitfield-Gabrielinn, Peter Williamsonoo,
Christian Windischbergerpp, Yu-Feng Zangqq, Hong-Ying Zhangii, F. Xavier Castellanosb,j, and Michael P. Milhamb,1
a
Department of Radiology, New Jersey Medical School, Newark, NJ 07103; bPhyllis Green and Randolph Cōwen Institute for Pediatric Neuroscience, New York
University Child Study Center, NYU Langone Medical Center, New York, NY 10016; cFMRIB Centre, Oxford University, Oxford OX3 9DU, UK; dHoward Hughes
Medical Institute, Harvard University, Cambridge, MA 02138; eSchool of Psychology, University of Wales, Bangor, UK; fDanish Research Centre for Magnetic
Resonance, Copenhagen University Hospital Hvidovre, Hvidovre, Denmark; gMood and Anxiety Disorders Program, National Institute of Mental Health/
National Institutes of Health, Department of Health and Human Services, Bethesda, MD 20892; hBehavioral Neuroscience Department, Oregon Health &
Science University, Portland, OR 97239; iDepartment of Diagnostic Radiology, Yale University School of Medicine, New Haven, CT 06511; jDivision of Clinical
Research, Nathan S. Kline Institute for Psychiatric Research, Orangeburg, NY 10962; kBiophysics Research Institute, Medical College of Wisconsin, Milwaukee,
WI 53226; lDepartment of Diagnostic Radiology, Oulu University Hospital, Oulu, Finland; mDonders Institute for Brain, Cognition, and Behavior, Center for
Neuroscience, Radboud University Nijmegen Medical Center, 6500 HB Nijmegen, The Netherlands; nBiophysics Research Institute, Medical College of
Wisconsin, Milwaukee, WI 53226; oInstitute of Neuroscience, National Yang-Ming University, Taiwan; pImaging Institute, The Cleveland Clinic, Cleveland, OH
44195; qBrain Imaging and Analysis Center, Duke University Medical Center, Durham, NC, 27710; rDepartment of Cognitive Neurology, Max Planck Institute
for Human Cognitive and Brain Sciences, 04103 Leipzig, Germany; sDepartment of Psychiatry and Department of Neurology, Emory University School of
Medicine, Atlanta, GA 30322; tCentre for Advanced Imaging, University of Queensland, Brisbane, Australia; uDepartment of Psychology, University of
Michigan, Ann Arbor, MI 48109; vLaboratory for Neurocognitive and Imaging Research, Kennedy Krieger Institute, Baltimore, MD, 21205; wDepartment of
Psychiatry, Oregon Health & Science University, Portland, OR 97239; xF.M. Kirby Research Center for Functional Brain Imaging, Kennedy Krieger Institute,
Baltimore, MD 21205; yFunctional MRI Laboratory, University of Michigan, Ann Arbor, MI 48109; zMcDonnell Center for Higher Brain Functions, Washington
University School of Medicine, St. Louis, MO 63110; aaDepartments of Neurology and Neuroradiology, Klinikum Rechts der Isar, Technische Universität
München, 81675 Munich, Germany; bbInstitute of Psychology and Department of Radiology, Leiden University Medical Center, Leiden University, Leiden,
The Netherlands; ccCenter for Brain Health and School of Behavioral and Brain Sciences, University of Texas at Dallas, Richardson, TX 75080; ddDepartment
of Neurology, Washington University School of Medicine, St. Louis, MO 63110; eeDepartment of Neurology, Charité Univesitaetsmedizin-Berlin, 10117
Berlin, Germany; ffSchool of Kinesiology, University of Michigan, Ann Arbor, MI 48109; ggDepartment of Psychiatry, University of Pittsburgh, Pittsburgh, PA
15213; hhDepartment of Psychiatry, Klinikum Rechts der Isar, Technische Universität München, D-81675 Munich, Germany; iiJiangsu Key Laboratory of
Molecular and Functional Imaging, Department of Radiology, Zhong-Da Hospital, Southeast University, Nanjing 210009, China; jjDepartment of Psychiatry,
Institute of Clinical Medicine and Department of Public Health Science, Institute of Health Science, University of Oulu, Oulu 90014, Finland; kkBerlin
NeuroImaging Center, 10099 Berlin, Germany; llDepartment of Psychiatry, Otto-von-Guericke University of Magdeburg, Magdeburg 39106, Germany;
mm
Laboratory for Higher Brain Function, Institute of Psychology, Chinese Academy of Sciences, Beijing 100864, China; nnDepartment of Brain and Cognitive
Sciences, Harvard-MIT Division of Health Sciences and Technology, Massachusetts Institute of Technology, Boston, MA 02139; ooDepartment of Psychiatry,
University of Western Ontario, London, ON N6A3H8, Canada; ppCenter for Medical Physics and Biomedical Engineering, Medical University of Vienna,
Vienna,
Austria; scripts
and qqState
Key Laboratory
Analysis
available
at of Cognitive Neuroscience and Learning, Beijing Normal University, Beijing 100875, China
http://fcon_1000.projects.nitrc.org
Edited*
by Marcus E. Raichle, Washington University, St. Louis, MO, and approved January 20, 2010 (received for review October 14, 2009)
#2 Site-­‐Related Varia<on is Very Real (and poten<ally addressable post-­‐hoc) Biswal et al., 2010
O
O
F
C.-G. Yan et al. / NeuroImage xxx (2013) xxx–xxx
Abraham et al., Varoquaux, in prep #3 Site-­‐Related Varia<on Goes Well Beyond Imaging Protocols Di Martino et al., 2010
Nooner et al., 2012
#4 Aggrega<on Promotes Replica<on Functional connectivity density mapping
Dardo Tomasia,1 and Nora D. Volkowa,b
a
National Institute on Alcohol Abuse and Alcoholism, Bethesda, MD 20892; and bNational Institute on Drug Abuse, Bethesda, MD 20892
Edited by Robert Desimone, Massachusetts Institute of Technology, Cambridge, MA, and approved April 21, 2010 (received for review February 4, 2010)
Brain networks with energy-efficient hubs might support the high
cognitive performance of humans and a better understanding of
their organization is likely of relevance for studying not only brain
development and plasticity but also neuropsychiatric disorders.
However, the distribution of hubs in the human brain is largely
unknown due to the high computational demands of comprehensive analytical methods. Here we propose a 103 times faster
method to map the distribution of the local functional connectivity
density (lFCD) in the human brain. The robustness of this method
was tested in 979 subjects from a large repository of MRI time
series collected in resting conditions. Consistently across research
sites, a region located in the posterior cingulate/ventral precuneus
(BA 23/31) was the area with the highest lFCD, which suggest that
this is the most prominent functional hub in the brain. In addition,
regions located in the inferior parietal cortex (BA 18) and cuneus
(BA 18) had high lFCD. The variability of this pattern across subjects
was <36% and within subjects was 12%. The power scaling of the
lFCD was consistent across research centers, suggesting that that
brain networks have a “scale-free” organization.
|
resting state functional MRI connectivity functional connectomes
mode networks scale-free networks consciousness
|
|
| default
T
overcome the limitations of seed-based approaches for the identification of hubs in the human brain, using resting-state functional
connectivity datasets. This ultrafast technique allows calculation of
individual functional connectivity maps with higher spatial resolution (≥3 mm isotropic) to take full advantage of the native resolution of the functional MRI datasets. The method is based on the
highly clustered organization of the brain (14). Specifically, to speed
up the computation of the number of functional connections (i.e.,
edges in graph theory), we restricted the temporal correlation
analysis to the local functional connectivity cluster.
Thus, we aimed to determine the location of the functional
connectivity hubs in the human brain by using data from the
“1000 Functional Connectomes Project” (17), which is a large
public database of resting-state time series that were collected
independently at 35 sites around the world (http://www.nitrc.org/
projects/fcon_1000/). We further aimed to evaluate the variability of the local functional connectivity density (lFCD) across
subjects and imaging parameters as well as its reproducibility
within and between subjects. We hypothesized that the lFCD
would have low within-subjects variability and that its spatial
distribution would be rather constant across research sites in the
world. We also hypothesized
probability
distribution
of (PC 1)
Fig. 3. Average that
spatialthe
distribution
of the first
principal component
across a
allpower
research sites
showing
the the
brain number
regions withofhigh
lFCD variance
the lFCD would have
scaling
with
func10–30%, radiological convention). Scatter plot shows lFCD
tional connections(red–yellow:
per node,
which is the main characteristic of
variance as a function of the principal components for each of the sessions of
the “scale-free” networks
(6),test–retest
ratherdataset.
than a Poisson distribution,
the New York
the landmark of random and “small-world” networks (2).
o support fast communication with minimal energy cost, cortical brain networks may have few nodes with dense local
clustering (hubs) and numerous nodes with an average low number
of connections (1–7). The energy-efficient regions (densely conrescaled lFCD was rather constant across these 979 subjects,
nected nodes) are thought to serve as the interconnection hubs, Results
regardless of differences in demographic variables between
Fig.
1 was
shows
the average
distribution
of the regions,
studies,
and
statistically
significant
in all gray matter
and neuropsychiatric diseases have been linked to abnormalities in Metaanalysis of FCDM.
even
whenacross
correcting
for
multiple
comparisons
atinthethis
voxel level
lFCD
in
the
human
brain
all
979
subjects
included
their configuration (8, 9). However, the investigation of hubs in the
with a conservative
familywise cingulate
error (FWE)
threshold PFWE <
study.
region
localized
within
the
posterior
cortex/venFig. 2. Spatial distribution of the average lFCD superimposed
on A
the middle
Fig. 1. Spatial distribution of the lFCD superimposed on axial MRI views of
brain
has(radiological
beenconvention).
hindered
by
the
cumbersome
0.05 (one-sample t test). Across subjects, the lFCDs in the possagital MRI plane computational
for all research sites (green labels) in this study (Table 2).
the
human brain
These maps
reflect
the average
tral precuneus (BAterior
23/31)
had the highest lFCD. Regions located in
FCDM parameters: T
number of functional connections per voxel (k) across 979 subjects from 19
cingulate/ventral precuneus and parietal hubs were 8.5 ±
requirements
of
comprehensive
analytical
methods. = 50 and T = 0.6.
research sites around the world. Green labels indicate the axial distance to
the cuneus, inferior parietal cortex, middle occipital, cingulate,
SNR
C
sub
mo
fol
Th
dit
sub
cen
Tes
FC
(N
lat
eta
ca
the
hig
the
for
fer
glo
Co
the
20
spa
ac
of
tha
12
RO
lFC
mo
to
ult
Developmental Effects S/P Volume Censoring and Mean FD Matching Fair et al.
Differentiating ADHD subtypes with rs-fcMRI
FIGURE A2 | Site-by-site histograms of Euclidean distance for functional connections identified in the whole group analysis that get stronger with
age and those that get weaker with age (FDR corrected) using procedure 8.
stay still, and fixate on a standard fixation-cross in the center of
the display.
Kennedy Krieger Institute
Participants were scanned using a 3.0 Tesla Philips scanner with
Fair et al., 2012
A comprehensive assessment of regional variation in the impact of head
micromovements on functional connectomics
Chao-Gan Yan a, b, c, Brian Cheung b, Clare Kelly c, Stan Colcombe a, R. Cameron Craddock b, d,
Adriana Di Martino c, Qingyang Li b, Xi-Nian Zuo e, F. Xavier Castellanos a, c, Michael P. Milham a, b,⁎
a
Nathan Kline Institute for Psychiatric Research, Orangeburg, NY, USA
Center for the Developing Brain, Child Mind Institute, New York, NY, USA
c
The Phyllis Green and Randolph Cowen Institute for Pediatric Neuroscience, New York University Child Study Center, New York, NY, USA
d
Virginia Tech Carilion Research Institute, Roanoke, VA, USA
e
Key Laboratory of Behavioral Science, Laboratory for Functional Connectome and Development, Magnetic Resonance Imaging Research Center, Institute of Psychology,
Chinese Academy of Sciences, Beijing, China
b
a r t i c l e
i n f o
Article history:
Accepted 5 March 2013
Available online 15 March 2013
Keywords:
Head motion correction
Resting-state fMRI
Voxel-wise movement
Test–retest reliability
Functional connectomics
a b s t r a c t
Functional connectomics is one of the most rapidly expanding areas of neuroimaging research. Yet, concerns remain regarding the use of resting-state fMRI (R-fMRI) to characterize inter-individual variation in
the functional connectome. In particular, recent findings that “micro” head movements can introduce artifactual inter-individual and group-related differences in R-fMRI metrics have raised concerns. Here, we first
build on prior demonstrations of regional variation in the magnitude of framewise displacements associated with a given head movement, by providing a comprehensive voxel-based examination of the impact of
motion on the BOLD signal (i.e., motion–BOLD relationships). Positive motion–BOLD relationships were
detected in primary and supplementary motor areas, particularly in low motion datasets. Negative motion–BOLD relationships were most prominent in prefrontal regions, and expanded throughout the brain
in high motion datasets (e.g., children). Scrubbing of volumes with FD > 0.2 effectively removed negative
but not positive correlations; these findings suggest that positive relationships may reflect neural origins
of motion while negative relationships are likely to originate from motion artifact. We also examined the
ability of motion correction strategies to eliminate artifactual differences related to motion among individuals and between groups for a broad array of voxel-wise R-fMRI metrics. Residual relationships between
motion and the examined R-fMRI metrics remained for all correction approaches, underscoring the need
to covary motion effects at the group-level. Notably, global signal regression reduced relationships between motion and inter-individual differences in correlation-based R-fMRI metrics; Z-standardization (mean-centering
and variance normalization) of subject-level maps for R-fMRI metrics prior to group-level analyses demonstrated
similar advantages. Finally, our test–retest (TRT) analyses revealed significant motion effects on TRT reliability
for R-fMRI metrics. Generally, motion compromised reliability of R-fMRI metrics, with the exception of those
based on frequency characteristics — particularly, amplitude of low frequency fluctuations (ALFF). The implications of our findings for decision-making regarding the assessment and correction of motion are discussed, as are
insights into potential differences among volume-based metrics of motion.
© 2013 Elsevier Inc. All rights reserved.
HarmonizaBon Needs Beyond AggregaBon: Preprocessing •  Lack of consensus regarding opBmal preprocessing (Craddock et al., 2013) –  Nuisance signal correcBon •  Physiologic signals •  MoBon signals •  Scanner arBfacts –  RegistraBon approaches/algorithms –  Temporal filtering –  Slice-­‐Bming Preprocessed INDI Data in the Cloud •  Available through S3 Bucket generously provided by AWS •  Raw INDI will be available soon h]p://preprocessed-­‐connectomes-­‐project.github.io/ Craddock et al., in prep
Analysis of Connectomes:
Pipeline for the Analysis
ctomes (C-PAC)
Craddock et al., in prep
PCP Quality Assessment Protocol •  SpaBal Measures –  Contrast to Noise RaBo –  Entropy Focus Criterion –  Foreground to Background Energy RaBo –  Smoothness (FWHM) –  % ArBfact Voxels –  Signal-­‐to-­‐Noise RaBo •  Temporal Measures –
–
–
–
–
Standardized DVARS Median distance index Mean FuncBonal Displacement # Voxels with FD > 0.2m % Voxels with FD > 0.2m h]p://preprocessed-­‐connectomes-­‐project.github.io/quality-­‐assessment-­‐protocol/ Craddock et al., in prep
Beyond AggregaBon: Synthesis •  AnalyBc VariaBon –  A growing plethora of approaches –  Even for same approach, marked variaBon can exist •  Seed-­‐based correlaBon approaches •  Parameter specificaBon –  ICA –  Graph theoreBcal approaches –  Cluster analysis •  ReporBng variaBons –  Inherent to all of fMRI That’s all Folks!