Peter Mondrup Rasmussen

Model sparsity and brain pattern interpretation of classification models in neuroimaging

Interest is increasing in applying discriminative multivariate analysis techniques to the analysis of functional neuroimaging data. Model interpretation is of great importance in the neuroimaging context, and is conventionally based on a brain map derived from the classification model. In this study we focus on the relative influence of model regularization parameter choices on both the model generalization, the reliability of the spatial patterns extracted from the classification model, and the ability of the resulting model to identify relevant brain networks defining the underlying neural encoding of the experiment. For a support vector machine, logistic regression and Fishers discriminant analysis we demonstrate that selection of model regularization parameters has a strong but consistent impact on the generalizability and both the reproducibility and interpretable sparsity of the models for both @?2 and @?1 regularization. Importantly, we illustrate a trade-off between model spatial reproducibility and prediction accuracy. We show that known parts of brain networks can be overlooked in pursuing maximization of classification accuracy alone with either @?2 and/or @?1 regularization. This supports the view that the quality of spatial patterns extracted from models cannot be assessed purely by focusing on prediction accuracy. Our results instead suggest that model regularization parameters must be carefully selected, so that the model and its visualization enhance our ability to interpret the brain.

Brain serotonin 2A receptor binding: Relations to body mass index, tobacco and alcohol use

Manipulations of the serotonin levels in the brain can affect impulsive behavior and influence our reactivity to conditioned reinforcers. Eating, tobacco smoking, and alcohol consumption are reinforcers that are influenced by serotonergic neurotransmission; serotonergic hypofunction leads to increased food and alcohol intake, and conversely, stimulation of the serotonergic system induces weight reduction and decreased food/alcohol intake as well as tobacco smoking. To investigate whether body weight, alcohol intake and tobacco smoking were related to the regulation of the cerebral serotonin 2A receptor (5-HT(2A)) in humans, we tested in 136 healthy human subjects if body mass index (BMI), degree of alcohol consumption and tobacco smoking was associated to the cerebral in vivo 5-HT(2A) receptor binding as measured with (18)F-altanserin PET. The subjects BMIs ranged from 18.4 to 42.8 (25.2+/-4.3) kg/m(2). Cerebral cortex 5-HT(2A) binding was significantly positively correlated to BMI, whereas no association between cortical 5-HT(2A) receptor binding and alcohol or tobacco use was detected. We suggest that our observation is driven by a lower central 5-HT level in overweight people, leading both to increased food intake and to a compensatory upregulation of cerebral 5-HT(2A) receptor density.

In vivo imaging of cerebral serotonin transporter and serotonin(2A) receptor binding in 3,4-methylenedioxymethamphetamine (MDMA or "ecstasy") and hallucinogen users.

CONTEXTnBoth hallucinogens and 3,4-methylenedioxymethamphetamine (MDMA or ecstasy) have direct agonistic effects on postsynaptic serotonin(2A) receptors, the key site for hallucinogenic actions. In addition, MDMA is a potent releaser and reuptake inhibitor of presynaptic serotonin.nnnOBJECTIVEnTo assess the differential effects of MDMA and hallucinogen use on cerebral serotonin transporter (SERT) and serotonin(2A) receptor binding.nnnDESIGNnA positron emission tomography study of 24 young adult drug users and 21 nonusing control participants performed with carbon 11 ((11)C)-labeled 3-amino-4-[2-[(di(methyl)amino)methyl]phenyl]sulfanylbenzonitrile (DASB) and fluorine 18 ((18)F)-labeled altanserin, respectively. Scans were performed in the user group after a minimum drug abstinence period of 11 days, and the group was subdivided into hallucinogen-preferring users (n = 10) and MDMA-preferring users (n = 14).nnnPARTICIPANTSnTwenty-four young adult users of MDMA and/or hallucinogenic drugs and 21 nonusing controls.nnnMAIN OUTCOME MEASURESnIn vivo cerebral SERT and serotonin(2A) receptor binding.nnnRESULTSnCompared with nonusers, MDMA-preferring users showed significant decreases in SERT nondisplaceable binding potential (neocortex, -56%; pallidostriatum, -19%; and amygdala, -32%); no significant changes were seen in hallucinogen-preferring users. Both cortical and pallidostriatal SERT nondisplaceable binding potential was negatively correlated with the number of lifetime MDMA exposures, and the time of abstinence from MDMA was positively correlated with subcortical, but not cortical, SERT binding. A small decrease in neocortical serotonin(2A) receptor binding in the serotonin(2A) receptor agonist users (both user groups) was also detected.nnnCONCLUSIONSnWe found evidence that MDMA but not hallucinogen use is associated with changes in the cerebral presynaptic serotonergic transmitter system. Because hallucinogenic drugs primarily have serotonin(2A) receptor agonistic actions, we conclude that the negative association between MDMA use and cerebral SERT binding is mediated through a direct presynaptic MDMA effect rather than by the serotonin(2A) agonistic effects of MDMA. Our cross-sectional data suggest that subcortical, but not cortical, recovery of SERT binding might take place after several months of MDMA abstinence.

Visualization of nonlinear kernel models in neuroimaging by sensitivity maps.

There is significant current interest in decoding mental states from neuroimages. In this context kernel methods, e.g., support vector machines (SVM) are frequently adopted to learn statistical relations between patterns of brain activation and experimental conditions. In this paper we focus on visualization of such nonlinear kernel models. Specifically, we investigate the sensitivity map as a technique for generation of global summary maps of kernel classification models. We illustrate the performance of the sensitivity map on functional magnetic resonance (fMRI) data based on visual stimuli. We show that the performance of linear models is reduced for certain scan labelings/categorizations in this data set, while the nonlinear models provide more flexibility. We show that the sensitivity map can be used to visualize nonlinear versions of kernel logistic regression, the kernel Fisher discriminant, and the SVM, and conclude that the sensitivity map is a versatile and computationally efficient tool for visualization of nonlinear kernel models in neuroimaging.

Obesity is associated with high serotonin 4 receptor availability in the brain reward circuitry

The neurobiology underlying obesity is not fully understood. The neurotransmitter serotonin (5-HT) is established as a satiety-generating signal, but its rewarding role in feeding is less well elucidated. From animal experiments there is now evidence that the 5-HT(4) receptor (5-HT(4)R) is involved in food intake, and that pharmacological or genetic manipulation of the receptor in reward-related brain areas alters food intake. Here, we used positron emission tomography in humans to examine the association between cerebral 5-HT(4)Rs and common obesity. We found in humans a strong positive association between body mass index and the 5-HT(4)R density bilaterally in the two reward ‘hot spots’ nucleus accumbens and ventral pallidum, and additionally in the left hippocampal region and orbitofrontal cortex. These findings suggest that the 5-HT(4)R is critically involved in reward circuits that regulate peoples food intake. They also suggest that pharmacological stimulation of the cerebral 5-HT(4)R may reduce reward-related overeating in humans.

PHYCAA: data-driven measurement and removal of physiological noise in BOLD fMRI.

The effects of physiological noise may significantly limit the reproducibility and accuracy of BOLD fMRI. However, physiological noise evidences a complex, undersampled temporal structure and is often non-orthogonal relative to the neuronally-linked BOLD response, which presents a significant challenge for identifying and removing such artifact. This paper presents a multivariate, data-driven method for the characterization and removal of physiological noise in fMRI data, termed PHYCAA (PHYsiological correction using Canonical Autocorrelation Analysis). The method identifies high frequency, autocorrelated physiological noise sources with reproducible spatial structure, using an adaptation of Canonical Correlation Analysis performed in a split-half resampling framework. The technique is able to identify physiological effects with vascular-linked spatial structure, and an intrinsic dimensionality that is task- and subject-dependent. We also demonstrate that increasing dimensionality of such physiological noise is correlated with increasing variability in externally-measured respiratory and cardiac processes. Using PHYCAA as a denoising technique significantly improves simulated signal detection with physiological noise, and real data-driven model prediction and reproducibility, for both block and event-related task designs. This is demonstrated compared to no physiological noise correction, and to the widely used RETROICOR (Glover et al., 2000) physiological denoising algorithm, which uses externally measured cardiac and respiration signals.

Nonlinear denoising and analysis of neuroimages with kernel principal component analysis and pre-image estimation

We investigate the use of kernel principal component analysis (PCA) and the inverse problem known as pre-image estimation in neuroimaging: i) We explore kernel PCA and pre-image estimation as a means for image denoising as part of the image preprocessing pipeline. Evaluation of the denoising procedure is performed within a data-driven split-half evaluation framework. ii) We introduce manifold navigation for exploration of a nonlinear data manifold, and illustrate how pre-image estimation can be used to generate brain maps in the continuum between experimentally defined brain states/classes. We base these illustrations on two fMRI BOLD data sets - one from a simple finger tapping experiment and the other from an experiment on object recognition in the ventral temporal lobe.

Prefrontal serotonin transporter availability is positively associated with the cortisol awakening response

UNLABELLEDnStress sensitivity and serotonergic neurotransmission interact, e.g. individuals carrying the low-expressing variants (S and LG) of the 5-HTTLPR promoter polymorphism of the serotonin transporter (SERT) gene are at higher risk for developing mood disorders when exposed to severe stress and display higher cortisol responses when exposed to psychosocial stressors relative to high expressing 5-HTTLPR variants. However, it is not clear how the relation between SERT and cortisol output is reflected in the adult brain. We investigated the relation between cortisol response to awakening (CAR) and SERT binding in brain regions considered relevant to modify the cortisol awakening response.nnnMETHODSnthirty-two healthy volunteers underwent in vivo SERT imaging with [(11)C]DASB-Positron Emission Tomography (PET), genotyping, and performed home-sampling of saliva to assess CAR.nnnRESULTSnCAR, defined as the area under curve with respect to increase from baseline, was positively coupled to prefrontal SERT binding (p=0.02), independent of adjustment for 5-HTTLPR genotype. Although S- and LG-allele carriers tended to show a larger CAR (p=0.07) than LA homozygous, 5-HTTLPR genotype did not modify the coupling between CAR and prefrontal SERT binding as tested by an interaction analysis (genotype×CAR).nnnCONCLUSIONnprefrontal SERT binding is positively associated with cortisol response to awakening. We speculate that in mentally healthy individuals prefrontal serotonergic neurotransmission may exert an inhibitory control on the cortisol awakening response.

Corrigendum to “Brain serotonin 2A receptor binding: Relations to body mass index, tobacco and alcohol use” [NeuroImage 46 (2009) 23–30]

a Neurobiology Research Unit, University Hospital Rigshospitalet, Copenhagen, Denmark b Center for Integrated Molecular Brain Imaging, Copenhagen, Denmark c Danish Center for Magnetic Resonance Imaging, Hvidovre University Hospital, Denmark d Department of Biostatistics, University of Copenhagen, Denmark e DTU Informatics, Technical University of Denmark, Lyngby, Denmark f PET and Cyclotron Unit, University Hospital Rigshospitalet, Copenhagen, Denmark

BredeQuery: Coordinate-Based Meta-analytic Search of Neuroscientific Literature from the SPM Environment

Large amounts of neuroimaging studies are collected and have chan-ged our view on human brain function. By integrating multiple studies in meta-analysis a more complete picture is emerging. Brain locations are usually reported as coordinates with reference to a specific brain atlas, thus some of the databases offer so-called coordinate-based searching to the users (e.g. Brede, BrainMap). For such search, the publications, which relate to the brain locations represented by the user coordinates, are retrieved. We present BredeQuery – a plugin for the widely used SPM data analytic pipeline. BredeQuery offers a direct link from SPM to the Brede Database coordinate-based search engine. BredeQuery is able to ‘grab’ brain location coordinates from the SPM windows and enter them as a query for the Brede Database. Moreover, results of the query can be displayed in a MATLAB window and/or exported directly to some popular bibliographic file formats (BibTeX, Reference Manager, etc).