Download Unsmoothed functional MRI of the human amygdala and bed

Author’s Accepted Manuscript
Unsmoothed functional MRI of the human
amygdala and bed nucleus of the stria terminalis
during processing of emotional faces
Ronald Sladky, Nicole Geissberger, Daniela M
Pfabigan, Christoph Kraus, Martin Tik, Michael
Woletz, Katharina Paul, Thomas Vanicek, Bastian
Auer, Georg S Kranz, Claus Lamm, Rupert
Lanzenberger, Christian Windischberger
PII:
DOI:
Reference:
www.elsevier.com
S1053-8119(16)30746-7
http://dx.doi.org/10.1016/j.neuroimage.2016.12.024
YNIMG13645
To appear in: NeuroImage
Received date: 30 July 2016
Revised date: 2 December 2016
Accepted date: 9 December 2016
Cite this article as: Ronald Sladky, Nicole Geissberger, Daniela M Pfabigan,
Christoph Kraus, Martin Tik, Michael Woletz, Katharina Paul, Thomas Vanicek,
Bastian Auer, Georg S Kranz, Claus Lamm, Rupert Lanzenberger and Christian
Windischberger, Unsmoothed functional MRI of the human amygdala and bed
nucleus of the stria terminalis during processing of emotional faces, NeuroImage,
http://dx.doi.org/10.1016/j.neuroimage.2016.12.024
This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal pertain.
Sladky et al., NeuroImage
Unsmoothed functional MRI of the human amygdala and bed
nucleus of the stria terminalis during processing of emotional faces
Ronald Sladky
1
1,2
1
3
4
1
, Nicole Geissberger , Daniela M Pfabigan , Christoph Kraus , Martin Tik , Michael
3
4
3
4
3
Woletz , Katharina Paul , Thomas Vanicek , Bastian Auer , Georg S Kranz , Claus Lamm , Rupert
4
Lanzenberger , Christian Windischberger
1
1,*
MR Center of Excellence, Center for Medical Physics and Biomedical Engineering, Medical
University of Vienna, Austria
2
Department of Psychiatry, Psychotherapy and Psychosomatics, Psychiatric Hospital, University of
Zürich, Switzerland
3
Social, Cognitive and Affective Neuroscience Unit, Department of Basic Psychological Research and
Research Methods, Faculty of Psychology, University of Vienna, Austria
4
Department of Psychiatry and Psychotherapy, Medical University of Vienna, Austria
*Correspondence to: Assoc. Prof. PD DI Dr. Christian Windischberger MR Center of
Excellence, Center for Medical Physics and Biomedical Engineering, Medical University of
Vienna, Austria, Lazarettgasse 14, 1090 Vienna, AUSTRIA. Tel: +43 1 40400 64630. Email:
[email protected]
1/27
Sladky et al., NeuroImage
Abstract
Functional neuroimaging of the human amygdala has been of great interest to uncover the
neural underpinnings of emotions, mood, motivation, social cognition, and decision making,
as well as their dysfunction in psychiatric disorders. Yet, several factors limit in vivo imaging
of amygdalar function, most importantly its location deep within the temporal lobe adjacent to
air-filled cavities that cause magnetic field inhomogeneities entailing signal dropouts.
Additionally, the amygdala and the extended amygdalar region consist of several
substructures, which have been assigned different functions and have important implications
for functional and effective connectivity studies. Here we show that high-resolution ultra-high
field fMRI at 7T can be used to overcome these fundamental challenges for acquisition and
can meet some of the demands posed by the complex neuroanatomy and -physiology in this
region. Utilizing the inherently high SNR, we use an optimized preprocessing and data
analysis strategy to demonstrate that imaging of the (extended) amygdala is highly reliable
and robust. Using unsmoothed single-subject data allowed us to differentiate brain activation
during processing of emotional faces in the central and basolateral amygdala and, for the
first time, in the bed nucleus of the stria terminalis (BNST), which is critically involved in the
neural mechanisms of anxiety and threat monitoring. We also provide a quantitative
assessment of single subject sensitivity, which is relevant for connectivity studies that rely on
time course extraction of functionally-defined volumes of interest.
Keywords (max 6).
Amygdala, BNST, faces, emotions, fMRI, 7 Tesla
2/27
Sladky et al., NeuroImage
1. Introduction
The amygdala is one of the most prominently researched brain structures in affective
neuroscience due to its pivotal role in the neural mechanisms of emotions and affectivity
(Fitzgerald et al., 2006; Ochsner et al., 2002), social cognition (Adolphs, 2010; Morris et al.,
2002), fear conditioning (Rogan et al., 1997; Tovote et al., 2015), and even complex decision
making in ambiguous and uncertain situations (Hsu et al., 2005; Ousdal et al., 2014). On a
more general level, the amygdala can be conceptualized as a fast-acting detector for
relevance and significance latent in the environment (Pessoa and Adolphs, 2010; Sander et
al., 2003). In addition, the amygdala is of utmost importance for clinical neuroscience and
psychiatry due to its observed dysfunctions in mood, anxiety, and a variety of psychiatric
disorders (Leppanen, 2006). Unfortunately, there are three major challenges associated with
in vivo investigations into human amygdalar functions and mechanisms: (1) its particular
localization in the ventral brain, (2) the functional segregation of amygdalar sub nuclei and
(3) the functional relevance of small subcortical structures surrounding the amygdala.
However, we argue that ultra-high field fMRI at 7 Tesla provides ways to overcome some of
the present limitations, offering novel research perspectives for improving our understanding
of the functions and dysfunctions of the human amygdala. The aim of this study is to
demonstrate 7T data quality and assess analysis results on single-subject and group level. In
addition, we aim at providing a benchmark for future high and ultra-high field studies of this
particular brain region to allow for feasibility evaluations and establishing an empirical
foundation for the selection of the appropriate imaging method and acquisition strategy.
In contrast to cortical regions, such as the visual or motor cortex, the amygdalae are located
deep within the temporal lobe and thus the application of non-invasive electrophysiological
methods is strongly limited. Among invasive approaches, intracranial EEG has provided
3/27
Sladky et al., NeuroImage
invaluable evidence for amygdala’s role during face perception (Huijgen et al., 2015; KrolakSalmon et al., 2004; Pourtois et al., 2010). The main drawback of these studies lies,
however, in the fact that they were performed only in small samples taken from clinical
populations undergoing brain surgery due to medical indications (e.g., epilepsy).
Generalization of these intracranial findings to an overall population is limited, as is the
application of this method to healthy subjects due to its invasive nature. As a fully noninvasive alternative, functional MRI enables high-resolution imaging of cortical and
subcortical regions and has become one of the most widely used methods for studying
human amygdala function in healthy subjects.
Nevertheless, fMRI in this brain region poses several methodological challenges. Blood
oxygen level dependent (BOLD) fMRI is an imaging method that is primarily based on the
contrast of susceptibility differences between oxygenated and deoxygenated hemoglobin,
which serves as a well-established proxy for clustered neural activity (Logothetis, 2002). The
amygdala and other ventral brain areas, however, are located in a part of the brain that is
strongly affected by inhomogeneities of the MR scanner’s static magnetic field B0 due to its
vicinity to air-filled cavities (i.e., sinuses) and bone structures. This leads to considerable
local image distortions and signal dropouts due to intra-voxel dephasing in gradient echo
planar imaging (Merboldt et al., 2001), as commonly used in fMRI. The negative effects of
intra-voxel dephasing can be efficiently compensated by reducing voxel sizes, i.e. increased
in-plane resolution and/or thinner slices (Morawetz et al., 2008; Robinson et al., 2004). A
number of fMRI studies have used such optimized acquisition protocols to obtain high quality
activation maps from the amygdala at a field strength of 3 Tesla (Derntl et al., 2009a; Derntl
et al., 2009b; Habel et al., 2007; Hahn et al., 2011; Karlsson et al., 2010; Morawetz et al.,
2010; Morawetz et al., 2016; Sladky et al., 2013; Sladky et al., 2012; Sladky et al., 2015a;
Sladky et al., 2015b). A second challenge for successful amygdala fMRI is the fact that
4/27
Sladky et al., NeuroImage
activation amplitudes in the amygdalae are considerably smaller than in their cortical
counterparts. As such, moving to ultra-high magnetic fields (7T and above) provides the
fundamental benefit that signal-to-noise ratio (SNR) scales approximately linear with field
strength when moving from 3T to 7T, as shown repeatedly (Triantafyllou et al., 2005; van der
Zwaag et al., 2009). There is, however, also an increase of artifacts originating from
magnetic field inhomogeneities and physiological noise (i.e., respiratory and cardiac signal)
(Hutton et al., 2011; Van de Moortele et al., 2002; Windischberger et al., 2010) and
sophisticated acquisition and analysis approaches are required to achieve high-quality fMRI
data. In a previous study, we have shown that the BOLD response in the amygdala is almost
linearly increased at 7 Tesla compared to 3 Tesla (left amygdala: +130%, right amygdala:
+80%) in a facial emotion discrimination task (Sladky et al., 2013). This increased overall
sensitivity even allows for detection of small signal differences in the amygdala such as those
induced by directed or averted gaze of fearful faces (van der Zwaag et al., 2012). Similar
benefits of ultra-high field fMRI have been observed for the visual cortex (Hoffmann et al.,
2009), the periaqueductal gray (Hahn et al., 2013), the hippocampus (Theysohn et al., 2013),
and ventral parts of the language network (Geissler et al., 2014).
The other major methodological challenge for non-invasive neuroscientific research on the
human amygdala is its anatomical and functional segregation into subareas or nuclei. Basal
and lateral amygdalar nuclei (BLA) receive neural input from unimodal and multimodal
sensory cortices, thalamus, and hippocampus and project back to multimodal sensory
cortices, the ventral striatum, and prefrontal regions (LeDoux, 2007). A second major nucleus
is the central amygdalar region (CeA) that integrates inputs from viscero-sensory cortex and
the sensory brain stem, which is involved in the processing of taste, pain, and visceral
information. Output of the CeA is considered to be important for physiological changes
associated with affective processes and includes arousal (via neuromodulatory systems),
5/27
Sladky et al., NeuroImage
freezing behavior (via the periaqueductal gray), activation of the sympathetic and
parasympathetic nervous systems and endocrine system (via the hypothalamus and nervus
vagus) (LeDoux, 2007). Additionally, small nuclei surrounding the amygdala play an
important role in amygdala regulation. For example, a small region of particular interest
within this extended amygdala network is the bed nucleus of the stria terminalis (BNST)
(Avery et al., 2014; Torrisi et al., 2015), which entertains direct anatomical connections to the
CeA. Its involvement in sustained anxiety as well as reward-processing and addiction (Avery
et al., 2016), which has also been observed in animal models, has important implications for
clinical neuroscience and psychiatric basic research in humans (Avery et al., 2016; Lebow
and Chen, 2016).
In humans, however, most fMRI findings are mainly at the level of the whole amygdalar
region rather than its individual nuclei (LeDoux, 2007). The reason for this is to a large
degree the limited attainable spatial resolution of fMRI at field strengths of 3T and below and,
more importantly, the limited signal-to-noise ratio (SNR). Moreover, a common preprocessing
technique to increase SNR of fMRI data is the use of spatial smoothing (e.g., using Gaussian
kernels with sizes of 6-12 mm FWHM), which additionally reduces spatial specificity
(Tabelow et al., 2009). Without the required spatial accuracy and sufficient SNR, it is not
possible to properly differentiate individual contributions of amygdala subareas or reliably
detect activation within extra amygdalar structures as small as the BNST.
The motivation of the present study was to investigate and quantify the potential of ultra-high
field, high-resolution fMRI of the human amygdala sub nuclei. We assessed the feasibility of
fMRI group analyses using unsmoothed single subject data for a complex task that involves
processing of emotional faces. Using independent anatomical masks, we quantified
individual signal changes within the BLA and CeA. Additionally, we explored whether the
acquisition setup presented herein also allows for the detection of task activation in the
6/27
Sladky et al., NeuroImage
BNST. Finally, we assessed the single-subject reliability of the observed task-dependent
activations within our volumes of interest, which could be crucial for feasibility evaluation and
power estimation required for future functional and effective connectivity studies in healthy
subjects and also clinical populations.
7/27
Sladky et al., NeuroImage
2. Participants and Methods
2.1. Study population
38 healthy volunteers (18f/20m, mean age ± SD: 27.5±6.8 years) were recruited for
participation in the fMRI study from the general public via billboard advertising. All subjects
underwent medical examination and an interview with an experienced psychiatrist including
the Structural Clinical Interview for DSM-IV (SCID) to rule out any psychiatric, neurological
and physical disorders. Further exclusion criteria were metal implants, past or present
substance abuse, intake of psychopharmacological medication within the last three months
and pregnancy (as assessed with a drug-urine and a urine human chorionic gonadotropin
pregnancy test, respectively). Written informed consent was obtained from each subject after
detailed explanation of the study protocol and subjects received reimbursement for their
participation. The study was approved by the ethics committee of the Medical University of
Vienna and the General Hospital of Vienna, Austria, and was performed in accordance with
the Declaration of Helsinki (1964) including current revisions.
2.2. 7T functional magnetic resonance imaging protocol
Measurements were performed on a MAGNETOM 7 Tesla whole-body MR scanner
(Siemens Medical, Erlangen, Germany) at the MR Centre of Excellence, Medical University
of Vienna. A 32-channel head coil (Nova Medical, USA) and the CMRR multiband EPI
sequences (Moeller et al., 2010) were used for data acquisition (245 whole-brain volumes,
repetition time TR=1.4s, echo time TE=23ms, flip angle 62°). Signal losses in the amygdala
region and orbitofrontal cortex were minimized by acquiring fMRI data with very high spatial
resolution: 78 slices with 1mm slice thickness (0.25mm gap), in plane resolution 1.5×1.5mm2
(Ganger et al., 2015).
2.3. Functional MRI task
8/27
Sladky et al., NeuroImage
Subjects were presented with a well-established emotion discrimination task (Hariri et al.,
2002). When presented with a triplet of emotional faces, subjects were instructed to indicate
by pressing a button whether the left or the right face matches the emotional valence of the
reference face presented in the middle. The presented faces were either neutral or
expressed anger, disgust, fear, happiness, sadness, surprise, and contempt. All trials were
self-paced. As a control condition, an object discrimination task was used, where subjects
had to match polygons that were superimposed on a skin-colored, circular background image
(Figure 1). All trials were presented in blocks of 20 second duration, alternating faces and
object blocks, with 20 second baseline (crosshair) blocks in-between. Each condition was
repeated four times, resulting in a total task duration of 5 minutes, 40 seconds. Each trial was
shown for at least 2 seconds and until a response was given via button press. Face stimuli
were obtained from the Radboud Faces Database (Langner et al., 2010) and object stimuli
were created in house.
Figure 1. Emotion and object discrimination task. Each task condition was presented four times in
alternating blocks (duration 20s). Subjects had to decide by button press if the left or right face/object
matches the one in the center.
2.4. Preprocessing and general linear model (GLM) analysis of fMRI data
9/27
Sladky et al., NeuroImage
Data were slice-timing corrected (FSL, http://fsl.fmrib.ox.ac.uk/fsl) (Sladky et al., 2011), biasfield corrected (ANTs, (Avants et al., 2011)), realigned (FSL), and non-linearly normalized to
MNI space (ANTs, final resolution 1.5×1.5×1.5 mm3) using a custom pre-processing pipeline.
Importantly, except the resampling to a marginally lower resolution during normalization, we
did not additionally smooth the data to preserve the spatial specificity in the main analysis.
Additionally, to demonstrate the effect of smoothing, VOI analyses were repeated using data
that has been smoothed with 3mm and 6mm FWHM Gaussian kernels. Data analyses were
performed in SPM (SPM12, build 6225, http://www.fil.ion.ucl.ac.uk/spm/software/spm12/)
using general linear model analysis. Single-subject GLM analyses included boxcar functions
convolved with SPM’s canonical hemodynamic response function to model the effects of
emotional faces and objects. Additionally, realignment parameter estimates and the first 6
principal components of white matter and ventricle time courses were added as nuisance
regressors (Sladky et al., 2013; Weissenbacher et al., 2009) to model for signal variability
unrelated to neural activity. Second-level random effects group analysis consisted of a onesample t-test on the individual contrasts between face and object discrimination
(FACES>OBJECTS). Significance threshold for the resulting statistical parametric map was
set to p<0.001 (voxel-wise threshold) and p<0.05 FWE cluster-level corrected, resulting in a
minimum cluster size of k=18 voxels.
2.5. Volume of interest analysis
To allow for an unbiased analysis, independent masks from the literature were used for
volume of interest (VOI) analysis. For definition of the BLA and CeA, we created masks
based on a high resolution amygdala parcellation by Tyszka and Pauli (2016), where the
amygdala was segmented into 9 sub nuclei. For the main analysis, the individual sub nuclei
were grouped into three clusters, given that some of the sub region masks could be too small
to be actually relevant for current fMRI studies, given the spatial extent of the BOLD
10/27
Sladky et al., NeuroImage
response and the typical voxel sizes that are used for image acquisition. In accordance with
the descriptions and definitions in their publication, we grouped the sub nuclei into: (1)
Basolateral amygdala (BLA) comprising lateral, basolateral, paralaminar, accessory basal
nuclei, (2) Central amygdala (CeA) comprising medial and cortical area, the nucleus of the
lateral olfactory tract and amygdalohippocampal area, as well as the periamygdaloid cortex,
and (3) the remaining amygdalar nuclei (OTH) comprising central, anterior amygdaloid area,
and amygdalostriatal transition area. However, to also report the detailed fMRI effects within
the small individual amygdalar sub structures, all 9 masks were used in a subsequent VOI
analysis.
To localize the BNST, we used an anatomically-defined mask by Torrisi et al. (2015).
11/27
Sladky et al., NeuroImage
3. Results
3.1. Behavioral data
On average, each subject solved 39.41 (SD=1.40) FACES and 39.49 (SD=1.52) OBJECTS
trials. Between conditions there was a moderate but significant difference in accuracy rates
(t(37)=3.28, Cohen’s d=0.53, p=0.006), 92.93% (SD=4.61%) for FACES and 95.23%
(SD=5.72%) for OBJECTS, and response times (t(37)=6.3, Cohen’s d=1.02, p<0.001),
1381ms (SD=323ms) for FACES and 1160ms (SD=287ms) for OBJECTS.
3.2. fMRI whole-brain data
For the target contrast FACES>OBJECTS highest activation was observed within the
extended amygdalar region (k=7579 voxels) with peak voxels in the left (T-20, -10, -16 =15.41)
and the right amygdala (T21, -8, -14 = 14.75). Note that voxel numbers are considerably high
compared to standard analyses because of the very high spatial resolution. The clusters with
the second and third strongest activation were found in the right fusiform gyrus (k=1169
voxels) with a peak at T48
-50 -30
= 7.03 and superior temporal lobe (k=4283 voxels) with a
peak at T48 -44 14 = 6.43. Large clustered activations were also observed in the left fusiform
gyrus and superior temporal cortex, additionally in other areas in the visual cortex, dorsoand ventrolateral PFC, as well as dorso- and ventromedial PFC. For a detailed report see
Table 1, Figure 2 for axial slices, and Figure 3 for coronal slices.
Importantly, we also observed significant brain activation within a cluster (k=60 voxel) with
peaks at T-4, -1, -2 =6.32 and T6, 0, -1 =5.56 that encompasses the BNST region, as defined by
the anatomical mask. Bilateral activations within the amygdala masks for BLA, CeA and OTH
were also reliably observable on group level (Figure 2, 3).
12/27
Sladky et al., NeuroImage
Figure 2. Mosaic of selected axial slices (z=-29.5 to +3.5 mm [MNI]) of the SPM t-map for FACES
> OBJECTS. Results are p<0.05 FWE cluster-level corrected (k=18 voxels minimum cluster size,
p<0.001 whole-brain threshold). Overlay of anatomical masks for basolateral amygdala (BLA, blue),
central amygdala (CeA, green), and other amygdalar regions (OTH, orange).
13/27
Sladky et al., NeuroImage
Figure 3. Detailed view of coronal slices of the SPM t-map for FACES > OBJECTS. Results are
p<0.05 FWE cluster-level corrected (k=18 voxels minimum cluster size, p<0.001 whole-brain
threshold). Overlay of anatomical masks for basolateral amygdala (BLA, blue), central amygdala (CeA,
green), other amygdalar nuclei (OTH, orange) and bed nucleus of the stria terminalis (BNST, yellow).
3.3. Volume of interest analyses
To quantify the effects of face in contrast to object discrimination within bilateral BLA, CeA,
OTH and BNST we conducted an additional VOI analysis. First, we assessed the average
percent signal change within the individual anatomical regions and found significant brain
activation in all VOIs. Median, interquartile range and distribution of the individual singlesubject mean percent signal changes are displayed in Figure 4 for unsmoothed data and
data that has been spatially smoothed with a Gaussian kernel of 3 mm and 6 mm FWHM.
The most prominent effect of smoothing was a reduction of effect sizes within all VOIs.
Using the individual amygdala sub nuclei masks, revealed that the strongest group activation
was found in the left and right paralaminar amygdala nucleus (BLV). After smoothing this
effect is no longer observed. A detailed visualization of the sub nuclear activation is provided
in the supplementary material (Supplementary Figure S1).
14/27
Sladky et al., NeuroImage
Figure 4. Violin plots of percent signal changes in left and right VOIs for FACES>OBJECTS.
Results are displayed for different levels of spatial smoothing (i.e., unsmoothed, 3 mm and 6 mm
FWHM). Complete distribution of single subject results (averaged within anatomical VOI) is plotted in
the background, white dot indicates median, and black bars represents the interquartile range. Left
and right basolateral amygdala (BLA, blue), central amygdala (CeA, green), other amygdalar nuclei
(OTH, orange), and bed nucleus of the stria terminalis (BNST, yellow). A more detailed report of the
activation within the amygdala sub nuclei is provided in Supplementary Figure S1.
In addition, methods that use functional localizers, such as a variety of real-time fMRI
neurofeedback setups and DCM and PPI data analysis strategies usually rely on extracting
15/27
Sladky et al., NeuroImage
time courses from activated voxels within a designated brain region. To provide this
information for future studies, the spatial extent of the activation was determined by
quantifying the percentage of significantly activated voxels within the VOIs per subject
relative to the thresholds of p<0.999, 0.500, 0.100, 0.050, 0.010, 0.005, and 0.001 (Figure 5).
The most robust activations were observed in the CeA, followed by BLA, and BNST. A
crucial question for methods that rely on extracting time courses from task-active voxels
(e.g., dynamic causal modeling) is the number of individual subjects from a study population
that exhibit significant activations within a given VOI. Using a common threshold of p<0.05,
significant voxels were observed in all 38 subjects for bilateral CeA, and the left BLA, in 37
subjects for the right BLA and the left BNST, and 36 subjects in the right BNST. With a more
liberal threshold of p<0.1, significant voxels for FACES>OBJECTS were observed in every
subject in all VOIs, except the right amygdala nuclei that are not part of the BLN or CMN
clusters (Figure 5).
Figure 5. Percentage of significantly activated voxels for basolateral amygdala (BLA), central
amygdala (CeA), other amygdalar nuclei (OTH), and bed nucleus of the stria terminalis (BNST).
Color coding indicates the percentage of significantly activated voxels within the respective VOI.
Single-subject sensitivity was consistently high within each VOI. E.g., for a typical threshold of p<0.05
significant voxels were detected in all BLA, CeA, and BNST VOIs in 36 out of 38 subjects, X-axis
shows typical single-subject thresholds and Y-axis indicates number of subjects.
16/27
Sladky et al., NeuroImage
4. Discussion
Using 7 Tesla ultra-high field MR we performed high-resolution functional MRI of small nuclei
within the (extended) amygdalar region in 38 healthy subjects performing an emotion
discrimination task. This is the first study that robustly detected activation in the amygdalar
region using unsmoothed fMRI data at 7T. Using this dataset with high spatial accuracy, we
were able to differentiate the BOLD response towards emotional faces (in contrast to object
discrimination) within the CeA and BLA. In addition, we show for the first time in humans
BNST activation that is related to emotional faces.
Beyond the ability to detect BOLD responses in small anatomical structures, analysis of
unsmoothed data entails two additional fundamental advantages that are particularly relevant
for functional and effective connectivity studies. First, from a methodological perspective,
unsmoothed data with sufficient SNR allows for a more accurate (e.g., anatomical or
functional) definition of task-active regions or voxels that can be used for time course
extraction. There is striking evidence that inappropriate definition of VOIs introduces biases
in the time series that are ‘extremely damaging to network estimation’ (Smith et al., 2011).
Importantly, these biases do not only degrade the sensitivity in connectivity analysis to some
degree. They also introduce the risk of incorrect estimations of connectivity strength,
direction, and network architecture.
Second, there are important anatomical differences between the nuclei in the amygdalar
region that also reflect fundamental functional differences as research investigating intraamygdala neural mechanisms in rodents has shown (LeDoux, 2007; Pitkanen et al., 1997).
Recent studies were able to uncover a microcircuit of GABA-erg neurons within CeA playing
an important role in encoding and gating of fear responses towards conditioned stimuli
(Ciocchi, 2010; Haubensak et al., 2010). These findings have important implications for
future research on the mechanisms of fear learning and reversal in humans.
17/27
Sladky et al., NeuroImage
The BNST, on the other hand, has shown to mediate sustained anxiety processes in rodents
(Davis et al., 1997; Sullivan et al., 2004; Waddell et al., 2006) and non-human primates
(Kalin et al., 2005). Increasing evidence for a similar role in humans is currently being
established (Avery et al., 2016) and supported by functional connectivity studies (Avery et al.,
2014; Torrisi et al., 2015).
BNST BOLD response to emotional faces was lower than in CeA and BLA. There are
methodological explanations such as the smaller size of this brain structure and that the
employed masks encompassed to a larger degree the lateral portions of the BNST (Torrisi et
al., 2015), while the focus of our activation was more located in the medial part. On the other
hand, the paradigm that we used is not particularly designed to induce anxiety, sustained
uncertainty, and long durations of threat, which are typically associated with BNST function.
However, the fact that we observed significant BNST activation indicates that emotional
faces can be used to induce BNST activation. This is important for clinical studies on, e.g.,
anxiety and affective disorders, where emotional faces are frequently used. Considering
BNST’s role in anxiety, we could expect increased activation towards emotional faces, e.g.,
in patients with social anxiety disorders. On the other hand, our results demonstrate the
feasibility to investigate task-related BOLD effects in the BNST in future studies where
induction of anxiety or longer phases of threat could be used as experimental paradigms.
Regarding
the
neurobiological
mechanisms
relevant
for
clinical
research,
BNST
catecholaminergic mechanisms, involving dopamine and norepinephrine, are assumed to be
influenced by a large variety of commonly administered antidepressants (Cadeddu et al.,
2014). Previous studies have already shown that, e.g., the selective serotonin reuptake
inhibitor citalopram can increase downregulation of the amygdala by the orbitofrontal cortex
(Sladky et al., 2015b), which is highly relevant, e.g., in social anxiety disorders where this
down regulation is significantly decreased compared to healthy participants (Sladky et al.,
18/27
Sladky et al., NeuroImage
2015a). By using high-resolution fMRI data, future studies on effective connectivity can
entertain more refined network models that allow for a more detailed estimation of neuronal
function and dysfunction. These models with improved biological plausibility would then allow
for a better translation of findings from animal models to non-invasive human (clinical)
neuroscience.
There are some limitations to the present study that can motivate future research directions.
First, the stimulation protocol that was used in this study is a very robust and reliable
paradigm to induce BOLD response in the amygdala region. However, it would be interesting
to use more complex designs to illuminate the individual functional contributions of the
different amygdala sub regions and the BNST in emotional face processing. Additionally, we
were not able to investigate the effects of explicit and implicit emotion regulation strategies
that were used by the participants using this simple paradigm. Next, including psychiatric
patients into the study population would be of high interest, as functional deviations that
might be characteristic for certain disorders can be expected in amygdala sub nuclei and
BNST. Then, regarding data acquisition, it has to be noted that we decided to use wholebrain coverage. Studies that focus only the amygdala and its sub nuclei could opt for partial
coverage in order to achieve even higher spatial and temporal resolution. Finally, there is an
increasing number of connectivity studies that demonstrate the benefits of high resolution 7T
data for the non-invasive in vivo investigation of neurobiological mechanisms in subcortical
regions (e.g., Metzger et al., 2013; Torrisi et al., 2015; Walter et al., 2008), which clearly
motivates additional studies on functional and effective connectivity within this network.
In summary, this study has demonstrated that ultra-high field fMRI in combination with
optimized data acquisition and preprocessing protocols allows for robust group analysis
based on unsmoothed, high-resolution single-subject data. We were able to discriminate the
individual contributions of CeA and BLA to the amygdalar BOLD signal in response to
19/27
Sladky et al., NeuroImage
emotional faces. Additionally, we observed face-related neural activation in the BNST, which
has important implications for future human in vivo studies on amygdala regulation, in
general, and specifically its role in anxiety and addiction disorders. Finally, we have
demonstrated that single-subject sensitivity of unsmoothed data is sufficient to allow for
reliable detection of task-activated voxels within our VOIs. These findings can be used as a
motivation for and justification of the feasibility of future time-series-based connectivity
studies within the extended amygdala.
Disclosure and Conflict of Interest
The authors declare no financial conflict of interest in the context of this study. This study
was supported by the research cluster MMI-CNS, funded by the Medical University of Vienna
and University of Vienna, Austria (FA103FC001) and by the CREAM project by the European
Commission under Grant Agreement (FP7-ICT-612022). This publication reflects the views
only of the authors, and the European Commission cannot be held responsible for any use
which may be made of the information contained therein.
20/27
Sladky et al., NeuroImage
References
Adolphs, R., 2010. What does the amygdala contribute to social cognition? Ann N Y Acad
Sci 1191, 42-61.
Avants, B.B., Tustison, N.J., Song, G., Cook, P.A., Klein, A., Gee, J.C., 2011. A reproducible
evaluation of ANTs similarity metric performance in brain image registration. Neuroimage 54,
2033-2044.
Avery, S.N., Clauss, J.A., Blackford, J.U., 2016. The Human BNST: Functional Role in
Anxiety and Addiction. Neuropsychopharmacology 41, 126-141.
Avery, S.N., Clauss, J.A., Winder, D.G., Woodward, N., Heckers, S., Blackford, J.U., 2014.
BNST neurocircuitry in humans. Neuroimage 91, 311-323.
Cadeddu, R., Ibba, M., Sadile, A., Carboni, E., 2014. Antidepressants share the ability to
increase catecholamine output in the bed nucleus of stria terminalis: a possible role in
antidepressant therapy? Psychopharmacology (Berl) 231, 1925-1933.
Ciocchi, S., 2010. Encoding of conditioned fear in central amygdala inhibitory circuits.
Nature.
Davis, M., Walker, D.L., Lee, Y., 1997. Amygdala and bed nucleus of the stria terminalis:
differential roles in fear and anxiety measured with the acoustic startle reflex. Philos Trans R
Soc Lond B Biol Sci 352, 1675-1687.
Derntl, B., Habel, U., Windischberger, C., Robinson, S., Kryspin-Exner, I., Gur, R.C., Moser,
E., 2009a. General and specific responsiveness of the amygdala during explicit emotion
recognition in females and males. Bmc Neuroscience 10, 91.
Derntl, B., Windischberger, C., Robinson, S., Kryspin-Exner, I., Gur, R.C., Moser, E., Habel,
U., 2009b. Amygdala activity to fear and anger in healthy young males is associated with
testosterone. Psychoneuroendocrinology 34, 687-693.
Fitzgerald, D.A., Angstadt, M., Jelsone, L.M., Nathan, P.J., Phan, K.L., 2006. Beyond threat:
amygdala reactivity across multiple expressions of facial affect. Neuroimage 30, 1441-1448.
Ganger, S., Hahn, A., Kublbock, M., Kranz, G.S., Spies, M., Vanicek, T., Seiger, R., Sladky,
R., Windischberger, C., Kasper, S., Lanzenberger, R., 2015. Comparison of continuously
acquired resting state and extracted analogues from active tasks. Hum Brain Mapp 36, 40534063.
Geissler, A., Matt, E., Fischmeister, F., Wurnig, M., Dymerska, B., Knosp, E., Feucht, M.,
Trattnig, S., Auff, E., Fitch, W.T., Robinson, S., Beisteiner, R., 2014. Differential functional
benefits of ultra highfield MR systems within the language network. Neuroimage 103, 163170.
Habel, U., Windischberger, C., Derntl, B., Robinson, S., Kryspin-Exner, I., Gur, R.C., Moser,
E., 2007. Amygdala activation and facial expressions: explicit emotion discrimination versus
implicit emotion processing. Neuropsychologia 45, 2369-2377.
Hahn, A., Kranz, G.S., Seidel, E.-M., Sladky, R., Kraus, C., Küblböck, M., Pfabigan, D.M.,
Hummer, A., Grahl, A., Ganger, S., 2013. Comparing neural response to painful electrical
stimulation with functional MRI at 3 and 7T. Neuroimage 82, 336-343.
Hahn, A., Stein, P., Windischberger, C., Weissenbacher, A., Spindelegger, C., Moser, E.,
Kasper, S., Lanzenberger, R., 2011. Reduced resting-state functional connectivity between
amygdala and orbitofrontal cortex in social anxiety disorder. Neuroimage 56, 881-889.
21/27
Sladky et al., NeuroImage
Hariri, A.R., Mattay, V.S., Tessitore, A., Kolachana, B., Fera, F., Goldman, D., Egan, M.F.,
Weinberger, D.R., 2002. Serotonin transporter genetic variation and the response of the
human amygdala. Science 297, 400-403.
Haubensak, W., Kunwar, P.S., Cai, H., Ciocchi, S., Wall, N.R., Ponnusamy, R., Biag, J.,
Dong, H.W., Deisseroth, K., Callaway, E.M., Fanselow, M.S., Luthi, A., Anderson, D.J., 2010.
Genetic dissection of an amygdala microcircuit that gates conditioned fear. Nature 468, 270276.
Hoffmann, M.B., Stadler, J., Kanowski, M., Speck, O., 2009. Retinotopic mapping of the
human visual cortex at a magnetic field strength of 7 T. Clinical Neurophysiology 120, 108116.
Hsu, M., Bhatt, M., Adolphs, R., Tranel, D., Camerer, C.F., 2005. Neural systems responding
to degrees of uncertainty in human decision-making. Science 310, 1680-1683.
Huijgen, J., Dinkelacker, V., Lachat, F., Yahia-Cherif, L., El Karoui, I., Lemarechal, J.D.,
Adam, C., Hugueville, L., George, N., 2015. Amygdala processing of social cues from faces:
an intracrebral EEG study. Soc Cogn Affect Neurosci 10, 1568-1576.
Hutton, C., Josephs, O., Stadler, J., Featherstone, E., Reid, A., Speck, O., Bernarding, J.,
Weiskopf, N., 2011. The impact of physiological noise correction on fMRI at 7 T. Neuroimage
57, 101-112.
Kalin, N.H., Shelton, S.E., Fox, A.S., Oakes, T.R., Davidson, R.J., 2005. Brain regions
associated with the expression and contextual regulation of anxiety in primates. Biol
Psychiatry 58, 796-804.
Karlsson, K.A.E., Windischberger, C., Gerstl, F., Mayr, W., Siegel, J.M., Moser, E., 2010.
Modulation of hypothalamus and amygdalar activation levels with stimulus valence.
Neuroimage 51, 324-328.
Krolak-Salmon, P., Henaff, M.A., Vighetto, A., Bertrand, O., Mauguiere, F., 2004. Early
amygdala reaction to fear spreading in occipital, temporal, and frontal cortex: A depth
electrode ERP study in human. Neuron 42, 665-676.
Langner, O., Dotsch, R., Bijlstra, G., Wigboldus, D., Hawk, S., Van Knippenberg, A., 2010.
Presentation and validation of the Radboud Faces Database. Cognition Emotion 24, 13771388.
Lebow, M.A., Chen, A., 2016. Overshadowed by the amygdala: the bed nucleus of the stria
terminalis emerges as key to psychiatric disorders. Mol Psychiatry 21, 450-463.
LeDoux, J., 2007. The amygdala. Curr Biol 17, R868-874.
Leppanen, J.M., 2006. Emotional information processing in mood disorders: a review of
behavioral and neuroimaging findings. Curr Opin Psychiatry 19, 34-39.
Logothetis, N.K., 2002. The neural basis of the blood-oxygen-level-dependent functional
magnetic resonance imaging signal. Philos Trans R Soc Lond B Biol Sci 357, 1003-1037.
Merboldt, K.D., Fransson, P., Bruhn, H., Frahm, J., 2001. Functional MRI of the human
amygdala? Neuroimage 14, 253-257.
Metzger, C.D., van der Werf, Y.D., Walter, M., 2013. Functional mapping of thalamic nuclei
and their integration into cortico-striatal-thalamo-cortical loops via ultra-high resolution
imaging - from animal anatomy to in vivo imaging in humans. Frontiers in Neuroscience 7.
Moeller, S., Yacoub, E., Olman, C.A., Auerbach, E., Strupp, J., Harel, N., Ugurbil, K., 2010.
Multiband Multislice GE-EPI at 7 Tesla, With 16-Fold Acceleration Using Partial Parallel
22/27
Sladky et al., NeuroImage
Imaging With Application to High Spatial and Temporal Whole-Brain FMRI. Magnetic
Resonance in Medicine 63, 1144-1153.
Morawetz, C., Baudewig, J., Treue, S., Dechent, P., 2010. Diverting attention suppresses
human amygdala responses to faces. Front Hum Neurosci 4, 226.
Morawetz, C., Bode, S., Baudewig, J., Kirilina, E., Heekeren, H.R., 2016. Changes in
Effective Connectivity Between Dorsal and Ventral Prefrontal Regions Moderate Emotion
Regulation. Cereb Cortex 26, 1923-1937.
Morawetz, C., Holz, P., Lange, C., Baudewig, J., Weniger, G., Irle, E., Dechent, P., 2008.
Improved functional mapping of the human amygdala using a standard functional magnetic
resonance imaging sequence with simple modifications. Magn Reson Imaging 26, 45-53.
Morris, J.S., DeBonis, M., Dolan, R.J., 2002. Human amygdala responses to fearful eyes.
Neuroimage 17, 214-222.
Ochsner, K.N., Bunge, S.A., Gross, J.J., Gabrieli, J.D., 2002. Rethinking feelings: an FMRI
study of the cognitive regulation of emotion. J Cogn Neurosci 14, 1215-1229.
Ousdal, O.T., Specht, K., Server, A., Andreassen, O.A., Dolan, R.J., Jensen, J., 2014. The
human amygdala encodes value and space during decision making. Neuroimage 101, 712719.
Pessoa, L., Adolphs, R., 2010. Emotion processing and the amygdala: from a 'low road' to
'many roads' of evaluating biological significance. Nat Rev Neurosci 11, 773-783.
Pitkanen, A., Savander, V., LeDoux, J.E., 1997. Organization of intra-amygdaloid circuitries
in the rat: an emerging framework for understanding functions of the amygdala. Trends
Neurosci 20, 517-523.
Pourtois, G., Spinelli, L., Seeck, M., Vuilleumier, P., 2010. Temporal precedence of emotion
over attention modulations in the lateral amygdala: Intracranial ERP evidence from a patient
with temporal lobe epilepsy. Cogn Affect Behav Neurosci 10, 83-93.
Robinson, S., Windischberger, C., Rauscher, A., Moser, E., 2004. Optimized 3 T EPI of the
amygdalae. Neuroimage 22, 203-210.
Rogan, M.T., Staubli, U.V., LeDoux, J.E., 1997. Fear conditioning induces associative longterm potentiation in the amygdala. Nature 390, 604-607.
Sander, D., Grafman, J., Zalla, T., 2003. The human amygdala: an evolved system for
relevance detection. Rev Neurosci 14, 303-316.
Sladky, R., Baldinger, P., Kranz, G.S., Trostl, J., Hoflich, A., Lanzenberger, R., Moser, E.,
Windischberger, C., 2013. High-resolution functional MRI of the human amygdala at 7 T. Eur
J Radiol 82, 728-733.
Sladky, R., Friston, K.J., Trostl, J., Cunnington, R., Moser, E., Windischberger, C., 2011.
Slice-timing effects and their correction in functional MRI. Neuroimage 58, 588-594.
Sladky, R., Hoflich, A., Atanelov, J., Kraus, C., Baldinger, P., Moser, E., Lanzenberger, R.,
Windischberger, C., 2012. Increased neural habituation in the amygdala and orbitofrontal
cortex in social anxiety disorder revealed by FMRI. PLoS One 7, e50050.
Sladky, R., Hoflich, A., Kublbock, M., Kraus, C., Baldinger, P., Moser, E., Lanzenberger, R.,
Windischberger, C., 2015a. Disrupted effective connectivity between the amygdala and
orbitofrontal cortex in social anxiety disorder during emotion discrimination revealed by
dynamic causal modeling for FMRI. Cereb Cortex 25, 895-903.
23/27
Sladky et al., NeuroImage
Sladky, R., Spies, M., Hoffmann, A., Kranz, G., Hummer, A., Gryglewski, G., Lanzenberger,
R., Windischberger, C., Kasper, S., 2015b. (S)-citalopram influences amygdala modulation in
healthy subjects: a randomized placebo-controlled double-blind fMRI study using dynamic
causal modeling. Neuroimage 108, 243-250.
Smith, S.M., Miller, K.L., Salimi-Khorshidi, G., Webster, M., Beckmann, C.F., Nichols, T.E.,
Ramsey, J.D., Woolrich, M.W., 2011. Network modelling methods for FMRI. Neuroimage 54,
875-891.
Sullivan, G.M., Apergis, J., Bush, D.E., Johnson, L.R., Hou, M., Ledoux, J.E., 2004. Lesions
in the bed nucleus of the stria terminalis disrupt corticosterone and freezing responses
elicited by a contextual but not by a specific cue-conditioned fear stimulus. Neuroscience
128, 7-14.
Tabelow, K., Piech, V., Polzehl, J., Voss, H.U., 2009. High-resolution fMRI: overcoming the
signal-to-noise problem. J Neurosci Methods 178, 357-365.
Theysohn, N., Qin, S., Maderwald, S., Poser, B.A., Theysohn, J.M., Ladd, M.E., Norris, D.G.,
Gizewski, E.R., Fernandez, G., Tendolkar, I., 2013. Memory-Related Hippocampal Activity
Can Be Measured Robustly Using fMRI at 7 Tesla. Journal of Neuroimaging 23, 445-451.
Torrisi, S., O'Connell, K., Davis, A., Reynolds, R., Balderston, N., Fudge, J.L., Grillon, C.,
Ernst, M., 2015. Resting State Connectivity of the Bed Nucleus of the Stria Terminalis at
Ultra-High Field. Hum Brain Mapp 36, 4076-4088.
Tovote, P., Fadok, J.P., Luthi, A., 2015. Neuronal circuits for fear and anxiety. Nat Rev
Neurosci 16, 317-331.
Triantafyllou, C., Hoge, R.D., Krueger, G., Wiggins, C.J., Potthast, a., Wiggins, G.C., Wald,
L.L., 2005. Comparison of physiological noise at 1.5 T, 3 T and 7 T and optimization of fMRI
acquisition parameters. Neuroimage 26, 243-250.
Tyszka, J.M., Pauli, W.M., 2016. In vivo delineation of subdivisions of the human amygdaloid
complex in a high-resolution group template. Hum Brain Mapp 37, 3979-3998.
Van de Moortele, P.F., Pfeuffer, J., Glover, G.H., Ugurbil, K., Hu, X., 2002. Respirationinduced B0 fluctuations and their spatial distribution in the human brain at 7 Tesla. Magn
Reson Med 47, 888-895.
van der Zwaag, W., Da Costa, S.E., Zürcher, N.R., Adams, R.B., Hadjikhani, N., 2012. A 7
tesla FMRI study of amygdala responses to fearful faces. Brain Topography 25, 125-128.
van der Zwaag, W., Francis, S., Head, K., Peters, A., Gowland, P., Morris, P., Bowtell, R.,
2009. fMRI at 1.5, 3 and 7 T: characterising BOLD signal changes. Neuroimage 47, 14251434.
Waddell, J., Morris, R.W., Bouton, M.E., 2006. Effects of bed nucleus of the stria terminalis
lesions on conditioned anxiety: aversive conditioning with long-duration conditional stimuli
and reinstatement of extinguished fear. Behav Neurosci 120, 324-336.
Walter, M., Stadler, J., Tempelmann, C., Speck, O., Northoff, G., 2008. High resolution fMRI
of subcortical regions during visual erotic stimulation at 7 T. MAGMA 21, 103-111.
Weissenbacher, A., Kasess, C., Gerstl, F., Lanzenberger, R., Moser, E., Windischberger, C.,
2009. Correlations and anticorrelations in resting-state functional connectivity MRI: a
quantitative comparison of preprocessing strategies. Neuroimage 47, 1408-1416.
Windischberger, C., Fischmeister, F.P.S., Schöpf, V., Sladky, R., Moser, E., 2010. Functional
magnetic resonance imaging with ultra-high fields. Der Radiologe 50, 144-151.
24/27
Sladky et al., NeuroImage
Table 1. SPM results of unsmoothed group analysis. Significance threshold for the resulting
statistical parametric map was set to p<0.001 voxel-wise threshold and p<0.05 FWE clusterlevel corrected, resulting in a minimum cluster size of k=18 voxels.
Cluster
Peak
MNI [mm]
p(FWE-corr)
p(unc.)
k [vx]
p(FWE-corr)
p(unc.)
T
x
y
<0.001
<0.001
7579
<0.001
<0.001
15.41
-20
-10
-16 Left Amygdala
<0.001
<0.001
14.75
21
-8
-14 Right Amygdala
<0.001
<0.001
10.39
15
-6
-25 Right Parahippocampal Gyrus
<0.001
<0.001
10.30
48
-50
-30 Right Fusiform Gyrus
<0.001
<0.001
10.03
45
-60
-24 Right Fusiform Gyrus
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
1169
4283
2670
3346
809
543
322
784
248
2650
674
z
Talairach-Tournoux Atlas / VOI masks
0.008
0.034
6.53
39
-64
-19 Right Fusiform Gyrus
<0.001
<0.001
8.82
48
-44
14
Right Superior Temporal Gyrus
0.001
<0.001
8.07
48
-68
18
Right Middle Temporal Gyrus
0.001
0.001
7.69
58
-46
16
Right Superior Temporal Gyrus
<0.001
<0.001
8.79
-2
-76
4
Left Cuneus
0.003
0.007
7.05
3
-68
12
Right PCC
0.004
0.007
7.02
-21
-58
-7
Left Parahippocampal Gyrus
<0.001
<0.001
8.48
-42
18
26
Left Middle Frontal Gyrus
0.001
0.001
7.74
-48
24
23
Left Inferior Frontal Gyrus
0.002
0.001
7.56
-33
11
28
Left Middle Frontal Gyrus
0.002
0.002
7.54
-14
65
30
Left Superior Frontal Gyrus
0.006
0.017
6.76
-8
50
50
Left Superior Frontal Gyrus
0.011
0.062
6.33
-8
59
38
Left Superior Frontal Gyrus
0.002
0.002
7.42
-50
-49
-30 Left Fusiform Gyrus
0.011
0.054
6.37
-48
-56
-26 Left Fusiform Gyrus
0.014
0.111
6.15
-45
-64
-19 Left Fusiform Gyrus
0.006
0.020
6.70
48
5
56
Right Middle Frontal Gyrus
0.014
0.113
6.14
38
2
42
Right Middle Frontal Gyrus
0.036
0.733
5.54
40
-1
65
Right Middle Frontal Gyrus
0.006
0.020
6.70
0
-60
35
Left Precuneus
0.027
0.415
5.72
4
-58
26
Right Precuneus
0.031
0.560
5.63
3
-60
46
Right Precuneus
0.006
0.022
6.66
6
20
62
Right Superior Frontal Gyrus
0.021
0.258
5.88
4
8
60
Right Superior Frontal Gyrus
0.105
1.000
4.88
4
18
71
Right Superior Frontal Gyrus
0.007
0.023
6.65
-64
-52
10
Left Superior Temporal Gyrus
0.008
0.029
6.57
-45
-56
20
Left Superior Temporal Gyrus
0.010
0.046
6.43
-48
-36
4
Left Superior Temporal Gyrus
0.010
0.050
6.40
3
41
25/27
-22 Right Orbital Gyru
Sladky et al., NeuroImage
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
43
259
60
0.013
0.093
6.20
-8
26
-24 Left Rectal Gyrus
0.014
0.123
6.11
-8
34
-19 Left Medial Frontal Gyrus
0.011
0.052
6.39
-8
-10
6
Left Thalamus
0.545
1.000
3.80
-3
-18
5
Left Thalamus
0.011
0.060
6.34
-38
5
44
Left Middle Frontal Gyrus
0.044
0.997
5.44
-45
10
52
Left Middle Frontal Gyrus
0.168
1.000
4.59
-36
0
36
Left Inferior Frontal Gyrus
0.011
0.064
6.32
-4
-1
-2
Left BNST
0.035
0.692
5.56
6
0
-1
Right BNST
0.115
1.000
4.82
3
-2
-12 Right Hypothalamus
0.039
<0.001
18
0.014
0.120
6.12
30
28
-2
<0.001
<0.001
85
0.023
0.313
5.81
-52
2
-22 Left Middle Temporal Gyrus
0.106
1.000
4.88
-52
-7
-18 Left Middle Temporal Gyrus
0.425
1.000
3.99
-54
0
-13 Left Middle Temporal Gyrus
Right Inferior Frontal Gyrus
<0.001
<0.001
76
0.023
0.326
5.80
0
-13
38
Left Cingulate Gyrus
<0.001
<0.001
422
0.024
0.348
5.78
-2
29
50
Left Superior Frontal Gyrus
0.030
0.496
5.67
-9
29
65
Left Superior Frontal Gyrus
0.040
0.872
5.48
-3
17
58
Left Superior Frontal Gyrus
0.028
0.451
5.70
8
46
40
Right Medial Frontal Gyrus
0.049
0.999
5.37
3
52
35
Right Superior Frontal Gyrus
0.236
1.000
4.37
4
60
35
Right Superior Frontal Gyrus
0.045
0.998
5.41
2
62
-12 Right Medial Frontal Gyrus
0.153
1.000
4.65
3
62
-20 Right Superior Frontal Gyrus
0.049
0.999
5.37
57
-52
-8
Right Middle Temporal Gyrus
0.242
1.000
4.35
64
-54
-6
Right Middle Temporal Gyrus
0.049
0.999
5.36
-44
18
-38 Left Superior Temporal Gyrus
0.201
1.000
4.47
-44
10
-40 Left Middle Temporal Gyrus
-36 Left Superior Temporal Gyrus
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
98
55
77
129
0.323
1.000
4.18
-38
14
<0.001
<0.001
63
0.088
1.000
5.00
15
-90
0.039
<0.001
18
0.110
1.000
4.85
0
-62
-36 Cerebellum
0.001
<0.001
31
0.118
1.000
4.81
18
-74
24
Right Precuneus
0.001
<0.001
28
0.143
1.000
4.69
-18
-73
22
Left Precuneus
0.646
1.000
3.67
-18
-79
30
Left Cuneus
-9
-94
0
Left Cuneus
5
Right Cuneus
<0.001
<0.001
33
0.149
1.000
4.66
0.005
<0.001
24
0.204
1.000
4.47
10
-14
8
Right Thalamus
0.001
<0.001
29
0.242
1.000
4.35
-18
-46
-6
Left Parahippocampal Gyrus
0.004
<0.001
25
0.435
1.000
3.97
16
-70
29
Right Precuneus
0.449
1.000
3.94
18
-80
34
Right Precuneus
26/27
Sladky et al., NeuroImage
Highlights (3-5. max 85 characters).

7 Tesla functional MRI of emotional face discrimination task in 38 healthy subjects

Unsmoothed single-subject data applicable for whole-brain group analysis

Differentiation of activation in central and basolateral amygdala

Significant effects in bed nucleus of stria terminalis, a key region for anxiety

Robust effects also on single-subject level, relevant for connectivity analyses
27/27