H

RRa

b

c

a

ARA

KM7VAE

1

1

bTsmirdt

istc

L

c

0d

European Journal of Radiology 82 (2013) 728– 733

Contents lists available at SciVerse ScienceDirect

European Journal of Radiology

journa l h o me pa ge: www.elsev ier .com/ locate /e j rad

igh-resolution functional MRI of the human amygdala at 7 T

onald Sladkya,b, Pia Baldingerc, Georg S. Kranzc, Jasmin Tröstl a,b, Anna Höflichc,upert Lanzenbergerc, Ewald Mosera,b, Christian Windischbergera,b,∗

MR Centre of Excellence, Medical University of Vienna, Lazarettgasse 14, 1090 Vienna, AustriaCenter for Medical Physics and Biomedical Engineering, Medical University of Vienna, Währinger Gürtel 18-20, 1090 Vienna, AustriaDepartment of Psychiatry and Psychotherapy, Medical University of Vienna, Währinger Gürtel 18-20, 1090 Vienna, Austria

r t i c l e i n f o

rticle history:eceived 22 August 2011ccepted 19 September 2011

eywords:agnetic resonance imaging (MRI)

Tentral brainmygdala

a b s t r a c t

Functional magnetic resonance imaging (fMRI) has become the primary non-invasive method for inves-tigating the human brain function. With an increasing number of ultra-high field MR systems worldwidepossibilities of higher spatial and temporal resolution in combination with increased sensitivity and speci-ficity are expected to advance detailed imaging of distinct cortical brain areas and subcortical structures.One target region of particular importance to applications in psychiatry and psychology is the amygdala.However, ultra-high field magnetic resonance imaging of these ventral brain regions is a challengingendeavor that requires particular methodological considerations. Ventral brain areas are particularlyprone to signal losses arising from strong magnetic field inhomogeneities along susceptibility borders.

motion discrimination In addition, physiological artifacts from respiration and cardiac action cause considerable fluctuationsin the MR signal. Here we show that, despite these challenges, fMRI data from the amygdala may beobtained with high temporal and spatial resolution combined with increased signal-to-noise ratio. Mapsof neural activation during a facial emotion discrimination paradigm at 7 T are presented and clearlyshow the gain in percental signal change compared to 3 T results, demonstrating the potential benefitsof ultra-high field functional MR imaging also in ventral brain areas.

. Introduction

.1. MR imaging at ultra-high fields

Within the last years, ultra-high field MR imaging of humanrain structure and function has become increasingly available.he high expectations from the application of scanners with fieldtrengths of 7 T and more [1] have triggered considerable invest-ents by research institutions across the world. Functional MR

maging (fMRI), employing fast imaging sequences, particularlyequires increased methodological considerations in hardwareesign, image acquisition and post-processing to fully benefit fromhe capacity of the latest advances in scanner engineering.

The most prominent benefit of high magnetic field strengthsn MR imaging is the approximately linear enhancement of image

ignal-to-noise ratio (iSNR) [2]. In addition, BOLD-weighted func-ional MRI application can also benefit from increased susceptibilityontrasts. A significant positive relationship between field strength

∗ Corresponding author at: MR Centre of Excellence, Medical University of Vienna,azarettgasse 14, 1090 Vienna, Austria. Tel.: +43 1 40400 6463.

E-mail addresses: ronald.sladky@meduniwien.ac.at (R. Sladky),hristian.windischberger@meduniwien.ac.at (C. Windischberger).

720-048X © 2011 Elsevier Ireland Ltd. oi:10.1016/j.ejrad.2011.09.025

Open access under CC BY-NC-ND license.

© 2011 Elsevier Ireland Ltd.

(comparison of 1.5, 3, and 7 T) and significant voxel counts, t-values,and amplitude of signal change within an a priori defined region ofinterest has also been reported [3].

The linear increase of iSNR, however, does not directly translateto a linear increase in BOLD sensitivity and temporal SNR (tSNR) at7 T. While the contributions of thermal noise are reduced, phys-iological influences (i.e. respiration and cardiac signal) mitigatethe measured temporal contrast induced by brain activation [2].At the same time, higher static magnetic field strengths (B0) leadto an increase in susceptibility-related field inhomogeneities andcause a reduction of T∗

2 relaxation times. The resulting shorter echotimes (e.g., cf. typical TE3 T = 42 ms vs. TE7 T = 23 ms) require fasterimage acquisition techniques, in particular the use of parallel imag-ing methods (e.g., GRAPPA, SENSE), and optimization of shimmingprocedures [4].

1.2. Functional MRI of the human amygdala

The ventral brain is particularly affected by both susceptibility-related effects and physiological artifacts. Field inhomogeneities

Open access under CC BY-NC-ND license.

surrounding the orbitofrontal, amygdalar, entorhinal, and inferiortemporal regions may cause severe signal dropout via intra-voxeldephasing effects, which make fMRI in these areas a challengingendeavor [5]. These areas are, however, of high relevance in the

nal of

pdpwe

obaatmc

saipBnotsttnp[orbd[mwf

vdanic

nf

Ff

R. Sladky et al. / European Jour

athogenesis of a variety of psychiatric diseases including anxietyisorders, major depression, and bipolar disorder. One region ofarticular interest is the amygdala. This almond-shaped structureithin the temporal lobe is considered an important hub in the

motion processing network.The amygdala consists of several nuclei distinguished by numer-

us connections to frontal cortical areas, the basal ganglia, therainstem and olfactory brain regions [6]. Due to its consider-ble link to other limbic structures such as the hippocampus, themygdala is mostly described as part of the limbic system. Besides,he amygdala receives various inputs from sensory cortices and

ediates a number of autonomous and motor functions via itsonnections to hypothalamus, thalamus and septal regions [7].

Current standard of knowledge from functional neuroimagingtudies outlines that the amygdala is essential to the evaluationnd processing of emotional stimuli, notably fear [8]. This theorys supported by the vast amount of fMRI studies using specificaradigms known to actuate amygdala reactivity. The amygdalarOLD response in healthy subjects has been shown to increase sig-ificantly in the presence of a strong trigger, such as fearful facesr negatively affected pictures [8,9]. Moreover, the magnitude ofhis signal is significantly increased in patients with major depres-ion and anxiety disorders, indicating that the hyperactivation ofhe amygdala may represent a trait marker of these disease pat-erns [10]. In addition, these findings are supported by moleculareuroimaging studies determining alterations in the amygdala inatients suffering from major depression and anxiety disorders11]. Furthermore, neuroscientific data suggest that the functionf the amygdala underlies an inhibitory control of frontal brainegions and that disturbances of this functional connectivity maye responsible for the development of several psychiatric disor-ers, such as posttraumatic stress disorder, social anxiety disorder12], bipolar disorder and depression. Nonetheless, to date the exact

echanisms underlying the pathogenesis of psychiatric diseases asell as the role of the amygdala in that context remain unclear and

urther investigations are certainly necessary.Particularly at very high magnetic field strengths, imaging the

entral brain region entails challenges. The amygdala and othereep-brain structures are located in close vicinity to bone andir-filled cavities (i.e. sinuses). Those areas are known for their vul-erability to susceptibility artifacts caused by local magnetic field

nhomogeneities [13] and physiologically induced artifactual signal

omponents [14,15].

On the other hand, the employment of optimized scan-ing protocols for high spatial resolution can compensate

or these challenges. The use of small voxel sizes (such as

ig. 1. Experimental paradigm. Facial emotion and object discrimination task blocks were

or 20 s to serve as a baseline condition. Each task block with an individual length of 20 s

Radiology 82 (2013) 728– 733 729

1.5 mm × 1.5 mm × 3 mm and smaller) has shown to reliably reducespin dephasing and signal drop-outs within the amygdala and otherparts of the limbic system as well as ventral areas of the orbito-frontal cortex [5]. Furthermore, the combination of high-resolutionscanning protocols and high field strength leads to increased rela-tive temporal SNR [2].

In the study presented we demonstrate that applications ofultra-high field fMRI are not limited to functional imaging ofprimary sensory and motor regions. Sequences and parametersoptimized for high-spatial resolution enable reliable signals fromventral brain areas that are highly reproducible at group level.

2. Methods

15 healthy volunteers (6 females, 9 males, mean age ± SD:29.54 ± 6.65 years) were recruited from the local community viabillboard announcements. Inclusion criteria for all subjects werephysical and mental health, no history in psychiatric or neurologicaldisorders, no past or current drug abuse, signed written informedconsent and age of 18–60 years. Volunteers were instructed torefrain from alcohol on the day of the measurement. All sub-jects were financially reimbursed for their participation. The studyprotocol was approved by the ethics committee of the MedicalUniversity of Vienna.

2.1. Emotion discrimination and object discrimination task

To actuate amygdala reactivity, subjects performed a facial emo-tion discrimination task (EDT) and an object discrimination task(ODT), adapted from the original paradigm introduced by Haririet al. [16]. Photos of emotional faces were obtained from the Rad-boud Faces Database [17]. The contours of regular polygons (3–7edges) were used for the ODT (Fig. 1).

The two discrimination blocks were alternately presented to thesubjects, with four 20 s blocks of each task condition and 20 s fix-ation cross baseline condition in-between, at the beginning andat the end of the run (total length: 340 s). In the EDT condition,participants were presented with a triplet of faces expressing oneof seven emotions (anger, disgust, fear, happiness, sadness, sur-prise, or calmness). They were instructed to select which of thetwo emotional faces presented left and right at the bottom of thescreen matches the target face at the top, by pressing either the

left or the right button of a MR-compatible response pad. In theODT condition the faces were replaced by contours of geometricalshapes, which were superimposed to a skin colored background.Subjects were told to respond after a waiting period of 2 s to ensure

alternately presented. Between task conditions, a black fixation cross was presentedwas repeated four times, yielding a total paradigm length of 340 s.

7 nal of

sur

lr

2

wEU3UsiFMdwp1e

2

(shhwFrtps

ebcrmdpTia

oawm

2

7ybGt1l

30 R. Sladky et al. / European Jour

ufficiently long exposure to the stimulus material. The next stim-lus was shown, when the subject pressed a button or failed toespond within 7 s.

Individual stimuli were projected to a semi-transparent screenocated at the back end of the scanner bore and presented in trueandomized order using Cogent 2000.

.2. Data acquisition

MRI measurements were performed on an ultra-high fieldhole-body 7 T MRI scanner (Magnetom 7 T, Siemens Medical,

rlangen, Germany) at the MR Centre of Excellence, Medicalniversity of Vienna, Austria. Subjects were scanned using a2-channel head array coil (Nova Medical, Wilmington, MA,SA). 245 whole-brain volumes (matrix size: 128 px × 128 px × 32

lices) were obtained at a repetition time of TR = 1.4 s employ-ng a single-shot echo planar imaging (EPI) sequence (TE = 23 ms,oV = 192 mm × 192 mm, 2 mm slice thickness). In order to reduceRI signal losses in ventral brain regions caused by intra-voxel

ephasing effects due to local magnetic field inhomogeneities,e applied an optimized protocol that provided high in-lane spatial resolution and thin slices with a voxel size of.5 mm × 1.5 mm × 2 mm. Parallel imaging (GRAPPA) was used tonable short TE required in ultra-high field fMRI.

.3. Pre-processing and data analysis

Acquired fMRI data were pre-processed and analyzed in SPM8FIL Group, UC London, UK). Preprocessing included correction forlice-timing differences [18], realignment to compensate for bulkead motion, normalization to standard MNI space using an in-ouse developed EPI template optimized for 7 T EPI imaging. Dataere spatially smoothed with an isotropic Gaussian kernel of 8 mm

WHM to reduce the impact of individual anatomical differences inandom-effects group analysis and account for the spatial distribu-ion of the BOLD effect. The estimated size of our target regions,articularly the amygdala, was the rationale for the size of themoothing kernel.

First-level single-subject analysis was conducted using the gen-ral linear model (GLM) framework implemented in SPM8. Twooxcar functions of the onsets and durations of the two taskonditions were convolved with SPM8’s canonical hemodynamicesponse function. These two separate regressors were designed toodel changes in hemodynamic responses during emotional face

iscrimination and object discrimination. Further, six realignmentarameters obtained from SPM8’s motion correction algorithm.ime courses from white-matter regions and ventricles were alsoncluded in the design matrix to reduce motion and physiologicalrtifacts.

Second-level group analysis was performed by conducting ane-sample t-test on the individuals’ contrasts between emotionnd object discrimination (i.e. EDT > ODT). Significance thresholdas set to p < 0.05 (FDR corrected for multiple comparisons) and ainimum cluster size of 100 voxels was chosen.

.4. Comparison of 3 T and 7 T data

For comparison, 15 healthy volunteers (different from the T group; 7 females, 8 females, mean age ± SD: 25.4 ± 3.4ears) participating in an ongoing study on a high field wholeody 3 T MRI scanner (Siemens Trio, Siemens Medical, Erlangen,

ermany), performed an emotion discrimination task comparable

o the design and setup described above (TR = 1.8 s; matrix size:28 px × 128 px × 20 slices). 3 T data were pre-processed and ana-

yzed using the aforementioned strategy and parameters.

Radiology 82 (2013) 728– 733

For direct comparison, one subject (female, 25 years) of the7 T study population underwent 3 T scanning with the identicalparadigm used at 7 T (TR = 1.8 s; matrix size: 128 px × 128 px × 23slices).

3. Results

High-quality 7 T BOLD fMRI datasets covering the whole cere-brum including amygdalar and other ventral regions were acquiredin all subjects. The result of the second-level group analysis is dis-played in Fig. 2 (p < 0.05 FDR corrected, n = 100 voxels minimumcluster size). Global maxima of the task-related contrast (emo-tion vs. object discrimination) were located in the left amygdala(T = 11.24) and right amygdala (T = 8.66). Other clusters were foundwithin the temporal lobe (fusiform gyrus, middle temporal gyrus),thalamus, insular cortex, posterior cingulate cortex and frontal lobe(bilaterally dorso-lateral prefrontal cortex, medio-frontal cortex).

On group level, amplitudes of task-related amygdalar BOLDsignal change were compared for 7 T and 3 T data and were sig-nificantly higher at 7 T: percent signal change in right amygdala:mean7 T = 0.70%, mean3 T = 0.38%, p = 0.02; percent signal change inleft amygdala: mean7 T = 0.88%, mean3 T = 0.38%, p = 0.01 (Fig. 3).

Fig. 4 shows EPI images of the participant measured at 7 T and3 T.

4. Discussion and outlook

In principle, functional imaging at 7 T allows for faster imageacquisition, higher spatial resolution and stronger BOLD-weighting.This increased contrast-to-noise ratio has been repeatedly demon-strated in fMRI studies at 7 T (cf. [3]). These studies, however, werelimited to non-ventral brain areas. Here we have shown that ultra-high field MR can be successfully applied to obtain functional MRimages with high temporal and spatial resolution not only in pri-mary sensory and motor regions, but also in ventral brain areasincluding the amygdala.

The methods necessary for high-quality 7 T fMRI in ventral brainregions, in particular the amygdala, need to target the two mostchallenging features: (1) signal loss from susceptibility-related fieldinhomogeneities and (2) strong temporal signal fluctuations dueto motion and physiological artifacts. We have used an approachbased on high spatial resolution to reduce signal losses due tointra-voxel spin dephasing. Smaller voxel reduce intra-voxel fieldinhomogeneities and thus lead to increased MR signals. At 3 T,higher spatial resolution implies prolonged image acquisition peri-ods and thus reduced temporal resolution. At 7 T, however, wecan benefit from the increased MR signal strength and use parallelimaging with high GRAPPA factors to speed up acquisition.

The second challenge for 7 T amygdala fMRI lies in the strong sig-nal fluctuations found in ventral brain regions. These fluctuationsarise from head motion and physiological artifacts.

The use of both high spatial resolution and nuisance regressorshas enabled us to increase fMRI sensitivity to a level where amyg-dalar activation can be easily observed. In fact, activation maximain our target contrast (emotion vs. object discrimination task) wereindeed located within the amygdala and other limbic regions. Agroup comparison of percental signal changes in the amygdala at3 T and 7 T has shown an increase of up to 100% for the 7 T fMRIstudy.

In the individual subject scanned on both systems the 7 T MRimages provide excellent contrast and, therefore, enabled proper

visual discrimination of structural details. The 3 T images were freeof geometric distortions. However, in comparison, the image con-trast was only sufficient to identify gross structures and did notallow for exact anatomical localization of the amygdala.

R. Sladky et al. / European Journal of Radiology 82 (2013) 728– 733 731

Fig. 2. fMRI group analysis of 7 T dataset. Second level group analysis (t-test for EDT > ODT) revealed significant activations within the left and right amygdalae (p < 0.05 FDRcorrected, n = 100 voxels minimum cluster size).

0 5 10 15 20 25

−0.50

−0.25

0.00

0.25

0.50

0.75

1.00

%7 Tesla 3 Tesl a

Emotion discrimination

Object discrimnation

Contrast of interest ± SEMEDT vs ODT

0 5 10 15 20 25 Seconds

−0.50

−0.25

0.00

0.25

0.50

0.75

1.00

Lef

t A

myg

dal

aR

igh

t A

myg

dal

a

Fig. 3. Task-related signal change for 15 subjects at 3 and 7 T. Mean and SEM of task-induced signal change in the amygdalae for the contrast of interest (EDT vs. ODT). Incomparison with 3 T, parameter estimates of activation change were significantly increased at 7 T.

732 R. Sladky et al. / European Journal of Radiology 82 (2013) 728– 733

F from

c In addi

eptffr

rnAfrDpttlshtiafla

omlmsssidfistf

qr

ig. 4. fMRI single subject analysis. These high-resolution EPI images were acquiredan be seen in close vicinity of the auditory canals and sinuses in both datasets.

ncreased in the 7 T images.

The amygdala has been classified as an important hub in themotion processing network and plays an important role in theathogenesis of psychiatric disorders. With a stimulus design thatriggers amygdalar activation and sequence parameters optimizedor ventral brain regions, high spatial accuracy of single-subjectunctional brain maps can be established, which contribute to aobust and reliable group-level analysis.

In addition to their important contribution to fundamentalesearch on brain function, current and future studies target theeuronal correlates of treatment response in various disorders.ssuming amygdala dysfunction represents a decisive promoting

actor for the development of psychiatric phenotypes, a treatmentesponse may be accompanied by a resolution of this disturbance.epressive patients show a greater amygdalar BOLD response in theresence of sad faces compared to healthy subjects, whereas the lat-er seem to display an increased signal in view of happy faces. Afterreatment with selective serotonin reuptake inhibitors, the first-ine treatment in major depression, the negative bias in affectedubjects seems to reverse and adjust to the amygdala activity inealthy controls [19]. Several fMRI studies corroborate this sugges-ion, demonstrating normalized activation in several brain regionsn major depression after antidepressant treatment, in particular

decrease of amygdala activation and an increase of activation inrontal areas [19,20]. These significant findings represent a majorandmark in the understanding of various disease patterns as wells the effects of commonly used medication.

It should be noted that these insights only represent a first partn the way of conceiving the human brain’s complexity; researchethods using the advantage of high spatial and temporal reso-

ution provided by ultra-high field MRI seem necessary to gainore precise data, in particular in view of small brain regions

uch as the amygdala with its differentiation in several nucleartructures. Recent developments in structural imaging techniquesuggest improved tissue contrast, ideal for segmentation. However,ts applicability to small structures in the ventral brain stronglyepends on the increased signal strength achieved with ultra-higheld MR imaging. In standard high-field MR scanning on 1.5–3 Tystems, on the contrary, the amygdala is imaged as a single, rela-ively homogenous portion of grey matter, therefore not allowing

or a distinction between individual clusters.

Furthermore, additional evaluation of ultra-high field MR datauality and comparison with 3 T datasets are indicated and cur-ently in preparation. Based on these assessments, new quality

a female subject (25 years) at 7 T (upper row) and 3 T (bottom row). Signal dropoutsition to the shorter acquisition time (TR = 1.4 s vs. TR = 1.8 s), image contrast was

estimations can be given to drive MR sequence development,improve coil designs, and optimize post-processing strategies.Given the increased contrasts at 7 T, novel insights into the tem-poral and spatial nature of the BOLD effect in the human brain canbe expected. Combining the recent methodological and hardwareimprovements would enable us to further investigate the centralrole of delicate structures, such as the amygdala, and their associ-ated networks in healthy brain function, as well as the pathogenesisand treatment of neuropsychiatric disorders.

Conflict of interest statement

The authors declare that there is no conflict of interest.

Acknowledgements

This research was funded by the Austrian Science Fund (FWFP23021) to Rupert Lanzenberger, the Austrian National Bank (ONBP12982) to Christian Windischberger, and via an unlimited researchgrant by Siemens Medical Solutions to Ewald Moser. This exper-iment was realised using Cogent 2000 developed by the Cogent2000 team at the FIL and the ICN and Cogent Graphics developedby John Romaya at the LON at the Wellcome Department of ImagingNeuroscience. We thank Andreas Hahn and Jan Losak for the imple-mentation and testing of the paradigm in Cogent and the clinicalstaff of the Department of Psychiatry and Psychotherapy.

References

[1] Moser E, Stahlberg F, Ladd M, Trattnig S. 7 Tesla MR from research to clinicalapplications? NMR in Biomedicine 2011;24.

[2] Triantafyllou C, Hoge RD, Krueger G, et al. Comparison of physiological noise at1.5 T, 3 T and 7 T and optimization of fMRI acquisition parameters. NeuroImage2005;26:243–50.

[3] van der Zwaag W, Francis S, Head K, et al. fMRI at 1.5, 3 and 7 T: characterisingBOLD signal changes. NeuroImage 2009;47:1425–34.

[4] Balteau E, Hutton C, Weiskopf N. Improved shimming for fMRI specificallyoptimizing the local BOLD sensitivity. NeuroImage 2010;49:327–36.

[5] Robinson S, Windischberger C, Rauscher A, Moser E. Optimized 3 T EPI of theamygdalae. NeuroImage 2004;22:203–10.

[6] Amunts K, Kedo O, Kindler M, et al. Cytoarchitectonic mapping of the human

amygdala, hippocampal region and entorhinal cortex: intersubject variabilityand probability maps. Anatomy and Embryology 2005;210:343–52.

[7] Karlsson KAE, Windischberger C, Gerstl F, Mayr W, Siegel JM, Moser E. Modu-lation of hypothalamus and amygdalar activation levels with stimulus valence.NeuroImage 2010;51:324–8.

nal of

[

[

[

[

[

[

[

[

[

[

R. Sladky et al. / European Jour

[8] Breiter HC, Etcoff NL, Whalen PJ, et al. Response and habituation of thehuman amygdala during visual processing of facial expression. Neuron1996;17:875–87.

[9] Derntl B, Windischberger C, Robinson S, et al. Amygdala activity to fear andanger in healthy young males is associated with testosterone. Psychoneuroen-docrinology 2009;34:687–93.

10] Phan KL, Fitzgerald DA, Nathan PJ, Tancer ME. Association between amygdalahyperactivity to harsh faces and severity of social anxiety in generalized socialphobia. Biological Psychiatry 2006;59:424–9.

11] Lanzenberger R, Mitterhauser M, Kranz GS, et al. Progesterone level predictsserotonin-1A receptor binding in the male human brain. Neuroendocrinology2011;94:84–8.

12] Hahn A, Stein P, Windischberger C, et al. Reduced resting-state functional con-nectivity between amygdala and orbitofrontal cortex in social anxiety disorder.

NeuroImage 2011;56:881–9.

13] Merboldt KD, Fransson P, Bruhn H, Frahm J. Functional MRI of the humanamygdala? NeuroImage 2001;14:253–7.

14] Biswal B, Deyoe E, Hyde J. Reduction of physiological fluctuations in fMRI usingdigital filters. Magnetic Resonance Imaging 1996;35:107–13.

[

Radiology 82 (2013) 728– 733 733

15] Windischberger C, Langenberger H, Sycha T, et al. On the origin of respira-tory artifacts in BOLD-EPI of the human brain. Magnetic Resonance Imaging2002;20:575–82.

16] Hariri A, Mattay V, Tessitore A, et al. Serotonin transporter genetic variationand the response of the human amygdala. Science 2002;297:400.

17] Langner O, Dotsch R, Bijlstra G, Wigboldus D, Hawk S, Van Knippenberg A. Pre-sentation and validation of the Radboud Faces Database. Cognition & Emotion2010;00:1–12.

18] Sladky R, Friston KJ, Tröstl J, Cunnington R, Moser E, Windischberger C.Slice-timing effects and their correction in functional MRI. NeuroImage2011.

19] Windischberger C, Lanzenberger R, Holik A, et al. Area-specific mod-ulation of neural activation comparing escitalopram and citalopramrevealed by pharmaco-fMRI: a randomized cross-over study. NeuroImage

2010;49:1161–70.

20] Delaveau P, Jabourian M, Lemogne C, Guionnet S, Bergouignan L, FossatiP. Brain effects of antidepressants in major depression: a meta-analysisof emotional processing studies. Journal of Affective Disorders 2011;130:66–74.