Post on 14-Jan-2016
Multivariate analyses in clinical populations: General
factors & neuroimaging
Joseph Callicott, MD
fMRI/MRI Summer Course 6/20/14
Introduction
The ‘Age of Big Data’ Lohr, “GOOD with numbers? Fascinated
by data? The sound you hear is opportunity knocking…” (NY Times, 2/22/2012)
We routinely collect ‘multimodal data’
E.g., mood rating scale and structural MRI Compile or compare, but typically without
multimodal analysisProjects classified as ‘geno-,’ ‘proteo-,’ or ‘pheno-’ already connotes ‘big data:’
Each fMRI image presents ~20K analyses GWAS model = strict correction for
multiple comparisons Current model = parallel
correlation/association per dataset Proposed model = multivariate approach
w/data reduction Simplified analyses Smaller statistical ‘cost’ Some current theoretical approaches become
testable’ (RDoC)
Outline
A Tale of Two Lectures:
I. Imaging genetics & schizophrenia
I. Relevant issues for clinical populations
II. Why imaging genetics?III. Imaging genetics 101IV. Multivariate analyses of fMRI:
within experimental dataset
II. General vs specific factors in dataI. gII. “i” : factor analytic solution of
general factors in fMRI task dataIII. Multivariate analyses of fMRI
redux: across experimental dataset
Issues of special interest to clinical studies…
BOLD fMRI in clinical populations:
BOLD fMRI is not, strictly speaking, a clinically informative measure No pathognomic findings, to date
Performance likely to differ fundamentally in tasks that HVs will perform near ceiling I.e., ~100% accuracy and faster RT than patients
BOLD fMRI in healthy subjects ‘predictive relationships’ between
behavior and BOLD implicit in design, perhaps not strongly correlated with group map activation
Genetic associations to BOLD (the bulk of the imaging genetics literature) do not necessarily connote a ‘real’ effect of a given polymorphism
Is the phenotype heritable? In past, twin or sibling studies Currently, within ‘only’ HV = GCTA (Visscher)
(Callicott et. al. 2000; Manoach et al., 2000; Callicott et al. 2003; others)
(Van Snellenberg et al. 2006)
So, then, why imaging genetics? Crass commercial message: o Simple plan, high impact
o Have an fMRI task in a relatively large sample?
o Healthy controls preferableo N > 40o Draw blood or swab cheek
o Genotyping at most resolutions fast & cheap*
o In SPM: ANCOVA or regression suffice and seem reasonably powered
COMT led fro
m primate to
human imaging, a
nd then to
drugs targetin
g cognitive
impairm
ents
Seriously, though, why imaging genetics?
o Few routes to neural mechanism using in vivo human datao Animal model o Drug studyo BOLD fMRI (MRSI, MEG, EEG)
o Genes do not code for mental illness, per seo Genes code for heritable aspects of brain function,
intermediate- or endo-phenotypeso Genetic risk for illnesses like schizophrenia is polygenic,
heterogeneouso Gene interact with each other and the environment
o BOLD fMRI, as an alternate metric of specific or general cognitive systems, offers ‘real world validation’ o Putative genetic mutations (including private mutations
(CNVs)o In spite of growing sample sizes, association studies risk
false positiveso RDoC domains and constructs
o If these do not correspond to brain systems we can map, then may be as doomed as DSM
Take home message…
Larger samples needed (GWAS), typically via collaboration across centers (ENIGMA)
BIG DATA is here Multivariate or non-
hypothesis-driven analyses offer the potential for novel, highly informative findings
CNV & cognition Very good software often
freely available: PLINK AFNI GingerALE-SLEUTH-MANGO R (many)
(Stefansson et al., Nature, 2014)
(*Visscher et al., 2010; Nan et al. 2012**; Postuma et al. 2002***; McGue & Bouchard, 1998#;^Burmeister et al., 2008)
Interest in imaging genetics predicated on heritability of phenotypes…
Callicott et al. Cereb Cortex 2000
Patients > Controls (N=13) (N=18)
Callicott et al. Am J Psychiatry 2003
Healthy Siblings > Controls (N=48) (N=33)
PFC BOLD during our Nback h2 = 0.4-0.5
Blokland et al. Biol Psych 2009, J Neurosci 2011; Koten et al. Science
2009
Finding genes for highly heritable, but complex diseases
affected person
unaffected
“nonpenetrant”
(Goldman et al. Nat Rev Genet 2006)
Remains difficult, even when n=100K1. Caused by many (100-1000s) of genes2. The effects of a mutation vary between people
Has all the genes (note this doesn’t mean
the exact same set)
May still carry some genes (like a parent of a
sick person)
Has all of the genes but is NOT sick for reasons
we can’t explain
Catechol O-methyltransferase (COMT): NIMH Intramural Success Story
(Apud et al. 2006)
(Egan et al. 2001) (Meyer-Lindenberg et al. 2006)
Functional impact COMT Val105/158Met
val/metrs4680
5’
Now validated at multiple levels:Animal models:• Reduced enzymatic activity• Altered synaptic dopamine
levels
Human data:• Reduced enzymatic activity
in vitro lymphoblastoid cell lines
• Altered transcription/reduced activity post mortem
• Altered D1 but not D2 receptor density in PFC
• PFC efficiency in BOLD fMRI
Combined: • Sex effects mostly in males
((Papaleo et al., 2014)
striatum
mammalianPFC
sibs
patients
controls
COMT Genotype
WC
ST
Pe
rse
ve
rati
ve
Err
ors
(t-
sc
ore
s)
30
35
40
45
50
55
60
v v v m m m
genotype effectF=5.41, df= 2, 449;
p<.005.
Executive cognition
Effect of rs4680 on frontal lobe function
(Egan et al PNAS 2001)
n = 218n = 181
n = 58
vv>vm>mm, SPM 99, p<.005
Physiological efficiency
Circa 2014: How have these findings held up?
Replicated but n’s ~20
14
BOLD phenotypes in simple association: COMT and PFC
(Mier, Kirsch, Meyer-Lindenberg; 2009)
15
BOLD phenotypes in simple association: 5-HTTP and Amygdala
(Munafo, Brown, and Hariri; 2007))
16
BOLD phenotypes in simple association: Power?
(Barnett et al, 2008)
1st generation imaging genetics: simple associationo Candidate genes
o KIBRA impaired memory & expressed in hippocampus (Papassotiropoulos et al., Science 2006)
o Replication in 3 independent populations in behavioral memory measures
o In 30 healthy subjects, KIBRA associated with reduced hippo activation
o Genetic mutations modeled in cell culture or animals
o Association based on disease GWAS (ZNF804A)o Esslinger et al. 2009 (Science)o Rasetti et al. (Arch Gen Psychiatry)
2nd generation imaging genetics: GWAS era
PFC neuronal function: ‘optimized’ by dopamine & GABA interactions
(Goldman-Rakic & Selemon 1997)
(Seamans et al., 2001)
2nd generation imaging genetics: Epistasis and pathways
(Straub et al., 2007)
2nd generation imaging genetics: Epistasis
COMT: V/V V/M
M/M
Bray Hap:
-/-
-/-
-/- +/- +/-
+/-
V/V V/M
M/M
V/V
V/M M/M
+/+
+/+
+/+
BO
LD
fM
RI
Lef
t D
LP
FC
(a
.u.)
COMT x Dysbindin interaction
o Epistasis (gene-gene interaction)o Initially based on candidate-by-candidate
o Buckholtz et al., Mol Psychiatry 2007o Data-driven (machine learning)
o Nicodemus et al., Hum Genet 2010o Now predicated on detailed cellular or animal modeling
o COMT x DTNBP1 (Papaleo et al., 2013)o DISC1 x NKCC1 (Kim et al., Cell 2012 & Callicott et al. J Clin
Invest 2013)
2nd generation imaging genetics: Translational neuroscience
22
More of the same (‘sophisticated univariate’)? Network/connectivity? Hypothesis-free?
Hypothesis-free pattern detection (random forest)
ICA/PCA/CPCA networks
Next generation imaging genetics?
Novel phenotypes (processing speed)
Sophisticated univariate: Imaging GWAS
o BOLD fMRI GWAS o Nback (n = 364)o Illumina 650K chip genotypingo Automated extraction of AAL ROIso First GWAS + using BOLD fMRI (Callicott, Spencer, et al., in prep)
As a heritable trait, BOLD fMRI phenotypes show other sensitivities…..o Long history within animal literature showing significant effects of environment
on brain structure & function o Beneficial effects of ‘enrichment’ (toys, limit isolation) (Hebb, Am J
Psychiatry, 1955) o fMRI during social stress task influenced by environment
o Urban upbringing or urbanicity linked to increased risk for mental illness (Van Os et al., Nature, 2010)
(Lederbogen et al., Nature, 2012)
Sophisticated univariate: Novel questions
As a heritable trait, BOLD fMRI phenotypes show other sensitivities…..
o fMRI during WM ( 3 cohorts (USA1 = 124; USA2 =92; Italy1=226 )o Sensitivity to childhood environment (Urbanicity) (Ihne et al., in submission)
Sophisticated univariate: Imaging G x E
o fMRI during WM ( 3 cohorts (USA1 = 124; USA2 =112; Italy1=226 )o Gene-by-environment interaction (COMT x Urbanicity) (Ihne et al., in submission)
Ihne et al., in preparation
Sophisticated univariate: Imaging G x E
27Constrained principle component analysis (CPCA) (David AA Baranger – Wash U)
http://www.nitrc.org/projects/fmricpca
• Todd Woodward and colleagues, University of British Columbia:• CPCA provides a “unified framework [for]… regression
analysis and principal component analysis .” • To identify functional systems using from singular-value
decomposition of BOLD time series, • These systems are imaged by constraining analyzed BOLD
signal from a particular interval of time against all other scans (i.e., all others are baseline)
Multivariate network analysis: ICA/PCA/CPCA banish ‘blob-ology’
28
CPCA
• Z or ‘activation’ matrix = individual time series for all subjects (rows) for all voxels in the brain (columns)
• Our standard SPM5 via XNAT first level processing of 0B alternating with 2B
• G or ‘design’ matrix = a model to predict BOLD signal changes (columns) over all fMRI scans (rows)
• SPM5 often uses a canonical hemodynamic response function (HRF) to deconvolve signal, fMRI-CPCA uses finite impulse response function (FIR)
http://www.nitrc.org/projects/fmricpca
29
CPCA
• N-back model not complicated:• Simply provide onset and
offset of 0B and 2B task epochs
• Components = extracted components represent networks
• Component loadings= loosely, correlation coefficients between component scores and BOLD signal that was predicted from imposed constraints (design)
http://www.nitrc.org/projects/fmricpca
30
• Identify and then display components using MRICon for anatomical localization (http://www.nitrc.org/projects/mricron)
• In this case, not really using estimated hemodynamics
• Rather, we wish to compare effect of diagnosis or genotype using component scores and predictor weights
• Predictor weight = contribution of G matrix to changes in components over the fMRI time series (~ correlation of component score and g)
CPCA
CPCA: Confusing Problematic Conflicting Agonizing
• Unspecified error required recalculation of component weights• Same networks found with addition of a fourth DMN
• Differentiates NC and SIBs from SCZ• No longer appears to be identifying intermediate phenotype
CPCA: Nback systems
Anti-task Networkresembles
cingulate from DMN+ hippocampus
WM NetworkDLPFC
+parietal
Motor system
Anti-task #2 resembles parietal
From DMN+ cerebellum
CPCA: Factors sensitive to disease, not genetics
p<0.05 p<0.05
• Unspecified error required recalculation of component weights• Same networks found with addition of a fourth DMN
• Differentiates NC and SIBs from SCZ• No longer appeared to identify intermediate phenotype
CPCA: Not particularly sensitive in general
• 420 HV • CPCA (2back) = 4 factors• Neuro- = 6 cognitive factors
• 2B as measured in lab• g estimates
CPCA F1 CPCA F2 CPCA F3 CPCA F42B accuracy 0.16 -0.21
2B accuracy (y) -0.082B RT (y) -0.11
F1 VerbalMemory 0.08F2 Nback
F3 VisualMemory 0.10 -0.14F4 ProcessingSpeed
F5 CardSortingF6 Span 0.09Little gBig g 0.10 (0.07)
Big data benefits reproducibility…
ENIGMA: first GWAS+ sMRI (Stein et al. 2009, 2010; Thompson et al., 2014)
Big data benefits reproducibility…
Heritability for novel phenes Replication on large scale
Outline
A Tale of Two Lectures:
I. Imaging genetics & schizophrenia
I. Why imaging genetics?II. Imaging genetics 101III. Multivariate analyses of
fMRI: within experimental dataset
II. General vs specific factors in dataI. gII. “i” : factor analytic solution of
general factors in fMRI task data
III. Multivariate analyses of fMRI redux: across experimental dataset
Are phenotypes independent?
Pearson’s Correlations
DSST
HV>SIB>SCZ [Tal:--47 7 39]
LDLFPC
DSST HV>SIB>SCZ [Tal: 20 -30 1]
Rhippo
DSST SCZ>SIB>HV
[Tal: -47 32 26] LBA9
DSST SCZ>SIB>HV [Tal: 36 6 53]
RBA6
DSST HV>SIB>SCZ [Tal:--47 7 39]
LDLFPC
Pearson Correlation 1 .052 .173** .174**
Sig. .412 .006 .006
N 249 249 249 249
DSST HV>SIB>SCZ [Tal: 20 -30 1]
Rhippo
Pearson Correlation .052 1 -.121 .046
Sig. .412 .056 .468
N 249 249 249 249
DSST SCZ>SIB>HV [Tal: -47 32
26] LBA9
Pearson Correlation .173** -.121 1 .194**
Sig. .006 .056 .002
N 249 249 249 249
DSST
SCZ>SIB>HV [Tal: 36 6 53]
RBA6
Pearson Correlation .174** .046 .194** 1
Sig. .006 .468 .002
N 249 249 249 249
Nback SCZ>SIB>HV [Tal: 21 16 60]
LBA6
Pearson Correlation .029 -.128* .152* .235**
Sig. .647 .044 .017 .000
N 249 249 249 249
Nback HV>SIB>SCZ [Tal: 32 -42 7]
Rhippo
Pearson Correlation .152* .034 -.042 .106
Sig. .016 .594 .511 .095
N 249 249 249 249
Are phenotypes independent?
ICC DSVT hypo Nback hyper NC (n=194)
Intraclass
Correlationa
95% Confidence Interval F Test with True Value 0
Lower Bound Upper Bound Value df1 df2 Sig
Single Measures .104b -.038 .243 1.233 189 189 .076
Average Measures .189c -.080 .391 1.233 189 189 .076
Are phenotypes independent?
The general cognitive factor (Spearman’s g)
(Dickinson et al., 2008)(Jensen, 1998)
Where is g?
‘Lesion maps’ from 241 patients w/ focal brain damage and g (Gläscher et al. PNAS 2010).
Barbey et al. (Brain 2012) found similar
results in 182 focal brain lesion patients
Various conceptual, functional and structural support for PFC and PAR (at minimum)
Is g associated with fMRI activation?
Nback (n= 161) higher g greater
efficiency
Replication (n= 582) higher g greater
efficiency
Exact overlap
Notes:1. Analysis: SPM5 multiple regression controlling
for age, sex2. 2B accuracy g (r = 0.3, p < 0.001)
Replication…cont’d
Replication 3 (n= 211)
Areas within replication exactly overlapping discovery…
discovery
Replication 5 (n= 306)
Replication 4 (n= 393)
Replication 4
Replication 3
Replication 5
DSVT v g
But…
Faces v g
MTL v g
g correlates with similar areas across 4 tasks in same 161 HVs
Nback (n= 161)
Notes:1. Analysis: SPM5 multiple regression controlling for
age, sex2. 161 with QC+ NB, MTL, Faces, DSVT3. NB as discovery ROI, others queried at p < 0.05
Is there a general solution for fMRI?
161 HVs with QC+ Nback, MTL (incidental encoding), Faces (response to aversive faces), and DSVT (processing speed)
Individual 1st level maps created for each task Sue Tong: automated script to extract parameter
estimates in Automated Anatomical Labeling (AAL) ROIs Mean fMRI ‘signal” transformed to Z score Factor analysis:
Principle component extraction Orthogonal and oblique rotations Factor scores estimated i (fMRI g) = sum of factor scores Comparison across task and against cognitive measures (big g)
iF1
Motor (R)Operculum (13)Cingulate (32)
SMA (R)Postcentral
Superior temporal gyrus (41,42)
Middle temporal gyrus (motion)
F2Insula
CaudatePutamen Pallidum
F3Cuneus (18)
Sup Occipital (19/7)Mid Occipital (39 &
19/37)Inf Occipital (V5/MT)
F4Cerebellum (8-
9)Cerebellar
vermis
F7SFG (8/9/6) (R)
MFG (R) IFG (44/45) (R)
Angular (R) (39)
F6SFG (8/9/6) (L)Medial SFG (8)Ant Cingulate
(24,32)Mid Cingulate
(24,31)F5
MFG (L)Sup Parietal (7)Inf Parietal (40)Supramarginal
(40)
fMRI (Nback) i(161 HVs, max likelihood extraction w/ varimax rotation, 60.1% total variance explained, goodness-
of-fit p < 1e-5)
.50
.41
.40 .42 .26
.42
.30
.13
.11
.13
.16
.10
fMRI (Nback) i
F7SFG (8/9/6) (R)
MFG (R) IFG (44/45) (R) Angular (R) (39)
SFG (R)
MFG (R) IFG (R)
Angular (R)
(161 HVs, max likelihood extraction w/ varimax rotation, 60.1% total variance explained, goodness-of-fit p < 1e-5)
.79
.90.59
.54
.7
.36
.43
.56
.52
.27
iF1
Motor (R)Operculum (13)Cingulate (32)
SMA (R)FPostcentral
Superior temporal gyrus (41,42)
Middle temporal gyrus (motion)
F2Insula
CaudatePutamen Pallidum
F3Cuneus (18)
Sup Occipital (19/7)Mid Occipital (39 &
19/37)Inf Occipital (V5/MT)
F4Cerebellum (8-
9)Cerebellar
vermis
F7SFG (8/9/6) (R)
MFG (R) IFG (44/45) (R)
Angular (R) (39)
F6SFG (8/9/6) (L)Medial SFG (8)Ant Cingulate
(24,32)Mid Cingulate
(24,31)F5
MFG (L)Sup Parietal (7)Inf Parietal (40)Supramarginal
(40)
fMRI (Nback) i
g 2B %C
g Pearson Correlation .302**
Sig. (1-tailed) .000
2B %C
Pearson Correlation .302**
Sig. (1-tailed) .000
F1(SMA-temporal) Pearson Correlation .045 .115
Sig. (1-tailed) .287 .074
F2 (basal ganglia) Pearson Correlation .062
Sig. (1-tailed) .217
F3 (visual) Pearson Correlation .045 .076
Sig. (1-tailed) .170
F4 (cerebellum( Pearson Correlation -.141* -.028
Sig. (1-tailed) .037 .361
F5 (L MFG-PAR) Pearson Correlation -.033 -.006
Sig. (1-tailed) .337 .470
F6 (SFG-ACING) Pearson Correlation .264** .188**
Sig. (1-tailed) .000 .009
F7 (R MFG-IFG-PAR) Pearson Correlation -.133* -.146*
Sig. (1-tailed) .046 .032
i Nback Pearson Correlation .042 .077
Sig. (1-tailed) .300 .166
**. Correlation is significant at the 0.01 level (1-tailed).
*. Correlation is significant at the 0.05 level (1-tailed).
i
F1Sup Occipital (7)
Mid Occipital (39 & 19/37)
Inf Occipital (V5/MT)Fusiform (37, FFA)
F2Calcarine (17/18)
Cuneus (18, V2)Lingual (19, V3)
F3PremotorMFG (R)
IFG (44/45) (R)Sup Parietal
(7)Inf Parietal (L)
(40)
F4Hippocampus
ParahippocampusAmygdala
Inf Temporal (L) (IT)
F6Cerebellum (1)Cerebellum (6)
F5SFG (8/9/6)
MFG (L)Medial SFG (8)
fMRI (Faces) i(161 HVs, max likelihood extraction w/ varimax rotation, 58.1 % total variance explained, goodness-
of-fit p < 1e-5)
.44
.40 .45 .40.39
.41
.10
.10
.10
.10.10
fMRI (Faces) i
g Faces RT
g Pearson Correlation -.259**
Sig. (2-tailed) .001
Faces RT Pearson Correlation -.259**
Sig. (2-tailed) .001
F1 (higher visual - PAR) Pearson Correlation -.206** .017
Sig. (2-tailed) .009 .834
F2 (lower visual) Pearson Correlation -.019 .141
Sig. (2-tailed) .811 .073
F3 (R MFG-IFG-PAR) Pearson Correlation -.095 -.034
Sig. (2-tailed) .228 .664
F4 (Amyg-Hippo) Pearson Correlation .125 -.013
Sig. (2-tailed) .113 .867
F5 (L MFG-SFG) Pearson Correlation .095 -.087
Sig. (2-tailed) .228 .273
F6 (Cerebellum) Pearson Correlation -.044 .011
Sig. (2-tailed) .584 .886
Faces i Pearson Correlation -.060 .015
Sig. (2-tailed) .453 .849
**. Correlation is significant at the 0.01 level (2-tailed).
i
F1Sup Occipital (7)
Mid Occipital (39 & 19/37)
Inf Occipital (V5/MT)Fusiform (37, FFA)
F2Calcarine (17/18)
Cuneus (18, V2)Lingual (19, V3)
F3PremotorMFG (R)
IFG (44/45) (R)
Sup Parietal (7)
Inf Parietal (L) (40)
F4Hippocampus
ParahippocampusAmygdala
Inf Temporal (L) (IT)
F6Cerebellum (1)Cerebellum (6)
F5SFG (8/9/6)
MFG (L)Medial SFG (8)
i
F1Orbitofrontal (47)Operculum (13)
InsulaSuparmarginal (L) (40)Sup temporal (41,42)
Mid temporal (TG)
F2SFG (8/9/6)
MFGIFG (44/45)Med SFG (8)
Ant Cing (24/32)Mid Cing (24/31)
F3HippocampusParahippocam
pus AmygdalaCaudate PutamenPallidumThalamus
F4Inf Occipital
Fusiform Sup Cerebellum
F6Calcarine (17/18)Cuneus (18, V2)Lingual (19, V3)
Sup Occipital (7)
F5Inf
Cerbellum
fMRI (DSVT) i(161 HVs, max likelihood extraction w/ varimax rotation, 51.6 % total variance explained, goodness-
of-fit p < 1e-5)
.40
.41 .42 .41.40
.41
.10
.10
.10.10
.10
fMRI (DSVT) i
g DSVT RT
g Pearson Correlation 1 .304**
Sig. (1-tailed) .000
DSVT RT Pearson Correlation .304** 1
Sig. (1-tailed) .000
F1(Orbito-Temp) Pearson Correlation -.057 -.008
Sig. (1-tailed) .237 .458
F2 (PFC-CING) Pearson Correlation .033 -.076
Sig. (1-tailed) .340 .170
F3 (Hippo-BG) Pearson Correlation .039 .076
Sig. (1-tailed) .310 .170
F4 (Lat Occ-Sup Cere) Pearson Correlation -.063 -.172*
Sig. (1-tailed) .214 .014
F5 (Inf Cerebellum) Pearson Correlation -.134* -.168*
Sig. (1-tailed) .045 .017
F6 (Visual-PAR) Pearson Correlation .122 .076
Sig. (1-tailed) .062 .168
G_med_dsvt Pearson Correlation -.025 -.111
Sig. (1-tailed) .378 .081
**. Correlation is significant at the 0.01 level (1-tailed).
*. Correlation is significant at the 0.05 level (1-tailed).
i
F1Orbitofrontal (47)Operculum (13)
InsulaSuparmarginal (L) (40)Sup temporal (41,42)
Mid temporal (TG)
F2SFG (8/9/6)
MFGIFG (44/45)Med SFG (8)
Ant Cing (24/32)Mid Cing (24/31)
F3HippocampusParahippocam
pus AmygdalaCaudate PutamenPallidum
Thalamus
F4Inf Occipital
Fusiform Sup Cerebellum
F6Calcarine (17/18)Cuneus (18, V2)Lingual (19, V3)
Sup Occipital (7)
F5Inf
Cerbellum
iF1
Calcarine (17/18)Cuneus (18, V2)Lingual (19, V3)Sup Occipital (7)
Mid Occipital (39 & 19/37)
Inf Occipital (V5/MT)Fusiform (37, FFA)
Inf Temporal (R) (IT)
F2Caudate PutamenPallidumThalamus
F3SFG (8/9/6)
MFGAngular (R) (39)
F4Insula
Sup Temporal (TG)
Mid Temporal (R) (TG)
F6Precentral (L)IFG (44/45)IFG (45/46)
Postcentral (L)Sup Parietal (L) (7)Inf Parietal (L) (40)Supramarginal (L)
(40)
F5Hippocampus
Parahippocampus
Inf Temporal (L) (IT)
fMRI (MTL) i(161 HVs, max likelihood extraction w/ varimax rotation, 51.6 % total variance explained, goodness-
of-fit p < 1e-5)
.40
.41 .42 .41.40
.41
.10
.10
.10.10
.10
fMRI (MTL) i
g MTL RT
g Pearson Correlation 1 -.071
Sig. (1-tailed) .186
MTL RT Pearson Correlation -.071 1
Sig. (1-tailed) .186
F1 (Visual) Pearson Correlation -.077 -.038
Sig. (1-tailed) .165 .317
F2 (Thalamus-BG) Pearson Correlation .065 -.031
Sig. (1-tailed) .205 .347
F3 (MFG-SFG-PAR) Pearson Correlation -.023 .051
Sig. (1-tailed) .388 .259
F4 (Insula-Temp) Pearson Correlation -.163* .091
Sig. (1-tailed) .020 .126
F5 (Hippo-Para) Pearson Correlation -.099 .249**
Sig. (1-tailed) .105 .001
F6 (IFG-L PAR) Pearson Correlation -.238** .120
Sig. (1-tailed) .001 .065
MTL i Pearson Correlation -.210** .174*
Sig. (1-tailed) .004 .014
*. Correlation is significant at the 0.05 level (1-tailed).
**. Correlation is significant at the 0.01 level (1-tailed).
iF1
Calcarine (17/18)Cuneus (18, V2)Lingual (19, V3)Sup Occipital (7)
Mid Occipital (39 & 19/37)Inf Occipital (V5/MT)Fusiform (37, FFA)
Inf Temporal (R) (IT)
F2Caudate PutamenPallidum
Thalamus
F3SFG (8/9/6)
MFGAngular (R) (39)
F4Insula
Sup Temporal (TG)Mid Temporal (R)
(TG)
F6Precentral (L)IFG (44/45)IFG (45/46)
Postcentral (L)Sup Parietal (L) (7)Inf Parietal (L) (40)Supramarginal (L)
(40)
F5Hippocampus
Parahippocampus
Inf Temporal (L) (IT)
An i by any other name…
g Faces i Nback i DSVT i MTL i
g Pearson
Correlation
1 -.060 .042 -.025 -.210**
Sig. (1-tailed) .226 .300 .378 .004
Faces i Pearson
Correlation
-.060 1 -.182* -.027 .014
Sig. (1-tailed) .226 .010 .365 .428
Nback i Pearson
Correlation
.042 -.182* 1 .056 -.049
Sig. (1-tailed) .300 .010 .241 .267
DSVT i Pearson
Correlation
-.025 -.027 .056 1 .124
Sig. (1-tailed) .378 .365 .241 .059
MTL i Pearson
Correlation
-.210** .014 -.049 .124 1
Sig. (1-tailed) .004 .428 .267 .059
**. Correlation is significant at the 0.01 level (1-tailed).
*. Correlation is significant at the 0.05 level (1-tailed).
Structural MRI
F1thick F2thick F3thick sMRI i
F1thick Pearson Correlation 1 -.580** .665** .668**
Sig. (2-tailed) .000 .000 .000
F2thick Pearson Correlation -.580** 1 -.386** .077
Sig. (2-tailed) .000 .000 .385
F3thick Pearson Correlation .665** -.386** 1 .802**
Sig. (2-tailed) .000 .000 .000
sMRI i Pearson Correlation .668** .077 .802** 1
Sig. (2-tailed) .000 .385 .000
Little_g Pearson Correlation -.037 -.072 -.031 -.094
Sig. (2-tailed) .679 .421 .727 .291
Big_g Pearson Correlation -.090 -.029 -.083 -.131
Sig. (2-tailed) .311 .749 .354 .140
Trans_Factor1_VerbalMemory Pearson Correlation .049 -.036 -.018 -.005
Sig. (2-tailed) .576 .682 .837 .955
Trans_Factor2_Nback Pearson Correlation .029 -.140 -.021 -.093
Sig. (2-tailed) .751 .123 .814 .309
Trans_Factor3_VisualMemory Pearson Correlation -.089 .044 -.077 -.076
Sig. (2-tailed) .339 .639 .411 .418
Trans_Factor4_ProcessingSpee
d
Pearson Correlation -.018 -.030 .027 -.015
Sig. (2-tailed) .842 .736 .761 .863
Trans_Factor5_CardSorting Pearson Correlation -.031 -.019 -.079 -.084
Sig. (2-tailed) .735 .835 .380 .350
Trans_Factor6_Span Pearson Correlation -.047 .007 -.019 -.037
Sig. (2-tailed) .600 .935 .833 .679
**. Correlation is significant at the 0.01 level (2-tailed).
(351 HVs, max likelihood extraction w/ varimax rotation, 48 % total variance explained, goodness-of-fit p < 1e-5)
Summary:
iF1
Calcarine (17/18)Cuneus (18, V2)Lingual (19, V3)Sup Occipital (7)
Mid Occipital (39 & 19/37)
Inf Occipital (V5/MT)Fusiform (37, FFA)
Inf Temporal (R) (IT)F2
Caudate PutamenPallidumThalamus
F3SFG (8/9/6)
MFGAngular (R)
(39)
F4Insula
Sup Temporal (TG)
Mid Temporal (R) (TG)
F6Precentral (L)IFG (44/45)IFG (45/46)
Postcentral (L)Sup Parietal (L)
(7)Inf Parietal (L)
(40)Supramarginal
(L) (40)F5
HippocampusParahippocam
pusInf Temporal
(L) (IT)
A Tale of Two Lectures:
I. Imaging genetics & schizophreniaI. Imaging genetics easier in an
age of data sharing and public databases
II. BOLD fMRI (IMHO) has never been about diagnosis = hello RDoC!
III. Multivariate analyses of fMRI: novel findings, novel questions
II. General vs specific factors in dataI. g inspires a straight-forward,
replicable multivariate analysis of fMRI (Factor analytic approach) (Dickinson et al., Biol Psych 2008; JAMA Psych 2014, numerous)
II. “i” : factor analytic solution of general factors in fMRI task data
III. Multivariate analyses of fMRI redux: I. Data reduction writ largeII. Replication across tasks, labs, designs?
Further musing…
Multimodal data, multimodal analysis
fMRI phenotypes are not independent Aspects within each task representing
individual ‘positive manifold’ Is heritability about this general shared
variance or specific task aspects?FMRI i as a data reduction method
Not g Complicated, but factor solution may be
informed by other data Structural MRI factor solution, no relationship
to g or other cognitive factors g holds special relationship to fMRI data
Test whether reduced factor structure more related to genes, other MRI, clinical measures
iF1
Calcarine (17/18)Cuneus (18, V2)Lingual (19, V3)Sup Occipital (7)
Mid Occipital (39 & 19/37)
Inf Occipital (V5/MT)Fusiform (37, FFA)
Inf Temporal (R) (IT)F2
Caudate PutamenPallidumThalamus
F3SFG (8/9/6)
MFGAngular (R)
(39)
F4Insula
Sup Temporal (TG)
Mid Temporal (R) (TG)
F6Precentral (L)IFG (44/45)IFG (45/46)
Postcentral (L)Sup Parietal (L)
(7)Inf Parietal (L)
(40)Supramarginal
(L) (40)F5
HippocampusParahippocam
pusInf Temporal
(L) (IT)
Thanks:Dwight Dickinson
Sue TongJessica Ihne
Karen Berman
Barbara SpencerGraham Baum
Morgan BartholomewAmanda Zheutlin
CTNB clinical staff