Archives 2010


Thursday, 16 December 2010, Visiting Speaker

Special time and location:  Thrursday, 4:00-5:00 pm,

Rue de Bugnon 21 (across the street from CHUV),  Salle de Seminaire 15, 6th floor.

Kerstin Preuschoff, 
Social and Neural Systems Lab
Institute for Empirical Research in Economics
University of Zurich

"The neurobiology of uncertainty during decision making and learning"

Throughout evolution, there has been nothing more certain than uncertainty. This talk will explore how behavior and neural networks have adapted to and reflect this omnipresence of uncertainty. Using different forms of uncertainty, such as risk and surprise, we link neural representations of uncertainty to models of reward based learning and decision making. We suggest that learning about uncertainty is supported by the noradrenergic system in a manner that parallels the dopaminergic reward prediction error hypothesis.




10 December 2010

1) Antonia Thelen, CHUV Science "The role of single-trial, episodic multisensory learning in unisensory object discrimination"

I will be presenting my Master thesis project. This project focused on how single-trial multisensory experiences can influence the ability to accurately discriminate image repetitions during a continuous recognition task. Previous studies have shown that pairing visual objects with their corresponding sounds can enhance subsequent visual discrimination, whereas pairing visual objects with an identical pure tone has been shown to impair subsequent visual discrimination compared with performance with objects only encountered visually. Despite their opposing polarity, these effects indicate incoming visual stimuli access multisensory memory traces established through single-trial learning. One open issue is the role of semantic versus episodic multisensory experiences, because prior work was confounded by pairing different visual objects with an identical pure tone. Here, we determined the role of episodic multisensory experiences by pairing (on their initial encounters) visual objects with meaningless, but unique sounds. Subjects discriminated initial from repeated presentations of images of common objects. To investigate this issue, we presented half of the initial presentations of images in a unisensory visual manner. Each of the remaining half of the images was paired on its initial presentation with a distinct but meaningless sound in a multisensory context. All repeated presentations were exclusively unisensory visual. Accuracy in recognition of repeated images was impaired for those that had been initially presented in a multisensory context. This decrement was dissociable from performance during initial image presentations, ruling out explanations in terms of attention or direct transfer from encoding to retrieval. Instead, the results indicate that the direction of the impact of single-trial multisensory memories on visual object discrimination is linked to the semantic versus episodic contingencies between the senses.




3 December 2010

1) Selma Aybek, CHUV Science "Conversion Disorder"
2) SPM course participants Methods "Short presentations on voxel-based morphometry (VBM)"

Selma Aybek:Conversion disorder is a frequent disabling disorder that presents with neurological symptoms thought to be a associated with psychological trauma. Psyhological models and neural correlates of conversion disorder, including fMRI data recently collected by our group, will be discussed.

SPM course: Today is the last day of a two-day workshop, offered by LREN, on "MRI:Statistical Parametric Mapping". Course participants will give short presentations on their data analysis using Voxel-Based Morphometry (VBM).




26 November 2010

1) Christian Wider, CHUV Science "Advances in genetics: implications for neurodegenerative diseases"
2) Tobias Kober, EPFL Methods "Prospective Motion Correction for Diffusion Imaging Using FID Navigators"


Christian Wider:Christian will discuss recent advances in the genetics of neurodegenerative diseases, focusing on parkinsonism and dystonia. The talk will emphasize how new technology influences the pace of discovery and how this transforms our understanding of the mechanisms involved in neurodegeneration. The high potential for translational research involving clinical characterization, imaging and neuropathology will also be discussed. Finally, implications for clinical practice / patient care will be emphasized.

Tobias Kober:Diffusion imaging is extremely susceptible to macroscopic motion. This is particularly critical in long acquisitions schemes, as they appear in diffusion tensor, spectrum and q-ball imaging, as well as clinical settings with young or uncooperative patients. The application of traditional registration-based correction methods is however problematic; the changing contrast between image volumes due to the varying diffusion encoding directions and weightings introduces a high uncertainty, in particular for diffusion weightings of b>500 s/mm2. Moreover, corrections a posteriori often do not consider the corresponding changes in the b-matrix, which may have a significant impact on the results. Free induction decay (FID) navigators have been shown to provide information about motion with no or negligible time penalty. In the presented project, we combined this navigator technique and a well-established registration method with the aim to prospectively or retrospectively correct motion in diffusion images of the head.




19 November 2010

1) Ulrike Toepel, CHUV Science "Spatio-temporal network dynamics for food evaluation depend on body weight and gender"
2) Fosco Bernasconi, CHUV Science "Attaining auditory temporal order judgement proficiency "



Ulrike Toepel: Women are more susceptible to eating disorders than men, suggesting that female and male brains might process food-related information differently. Moreover, weight status as measured by body-mass-index (BMI) has been shown to modulate food-related brain responses using hemodynamic imaging. Hitherto, the spatio-temporal brain dynamics of gender- and weight-wise modulations in perceiving food images have not been explored. For these purposes, we analyzed visual evoked potentials (VEPs) while normal-weighted men and women categorized photographs of energy-dense foods and non-food kitchen utensils. The VEP analyses revealed gender-wise VEP differences starting ~170ms after image onset. A cluster analysis of the VEP topography identified the 170-215ms post-stimulus onset interval as differing as a function of both gender and image category. Different VEP topographies accounted for responses to foods in women and men. No difference between the genders was observed for responses to non-food images. Over the same period, estimations of the neural generator activity by means of a distributed linear inverse solution (LAURA) showed that neural source strength was negatively correlated with BMI during food viewing in women only. In particular, these correlations indicating decreased neural source activity with increasing weight were apparent in ventral prefrontal brain regions known to be implicated in reward valuation and control over food intake. That is, the observed temporal and spatial brain response modulations to food viewing and discrimination likely indicate functional variations between genders in terms of food intake control and valuation, presenting a potential link to women's greater susceptibility to eating disorders.

Fosco Bernasconi: The building up of a coherent representation of the auditory environment requires an accurate integration of the order of stimuli occurrence within rapidly varying auditory streams. Our studies aim to investigate the neural underpinnings of auditory temporal order judgment (aTOJ). In order to assess this question, we recorded EEG while participants where trained in an aTOJ task in which they had to discriminate which of two spatially distinct auditory stimuli, separated by short delay, occurred first. In a first study, in order to identify the neural underpinning of successful aTOJ, we contrasted EEG responses to the pairs of sounds as a function of aTOJ accuracy (i.e. accurately vs inaccurately perceived trials). AEPs to sound pairs modulates topographically at ca. 40ms post-sound one onset, indicating the engagement of distinct configurations of brain networks during early processing stages of the auditory stimuli. Source estimations revealed that accurate and inaccurate responses were associated with bilateral posterior sylvian regions (PSR) activity. However, activity within left, but not right, PSR predicted behavioral performance. In a second study, in order to identify the neural mechanisms underlying aTOJ improvement, we contrasted responses recorded at the beginning vs. end of the aTOJ training task. The effect of training revealed corresponding patterns of results as for the accurate vs inaccurate contrast. AEP modulated topographically as a function of training blocks at ca 40 ms. Source estimations revealed that behavioral improvement in aTOJ accuracy was accompanied by an increase in the lateralization of an initially bilateral PSR responses to left-hemisphere dominance at the end of training. Finally, our two study reveal that accurate aTOJ as well as aTOJ improvement are supported by a functional decoupling between left and right PSR.




12 November 2010

1) Micah Murray, CHUV Science "Spatio-temporal brain dynamics of perceptual filling-in"
2) Eleonora Fornari Protocol "Changes in availability of CHUV 3T MRI
Trio for research"


Micah Murray:
The human visual system is adept at overcoming both quantitative and qualitative variation in visual scenes, such that object recognition is possible despite degraded visual conditions and impediments that produce discontinuous or absent boundaries within or between objects. Experimentally, stimuli producing illusory contours (ICs) have been used to mimic these conditions. Despite extensive research in humans and animals, controversy persists regarding the neurophysiologic mechanisms of IC sensitivity. Three general models can be distinguished. One favors effects within lower-tier cortices, V1/V2, mediated by feed-forward inputs and/or long-range horizontal interactions. Another situates IC sensitivity within higher-tier cortices, principally lateral-occipital cortex (LOC), with feedback effects in V1/V2. Still others postulate that the LOC is sensitive to salient regions demarcated by the inducing stimuli, whereas effects within V1/V2 reflect specifically IC sensitivity. This talk will present an overview of our work aimed at resolving discordances between these models and more generally at understanding perceptual filling-in.

Eleonora Fornari:
She will share with us the latest updates on CHUV 3T MRI Trio availability for research and procedures for reserving scan time.




5 November 2010

1) Dominica Bueti, Santa Lucia Foundation, Rome Science "The sensory representation of time"


Along with space, time is one of the fundamental dimensions for the perception of the external world; our ability to effectively respond to the dynamic events in our environment is strongly dependent on the accuracy with which we represent their temporal properties. Despite the importance of the temporal dimension the understanding of the neural mechanisms of temporal computations is relatively poor . Neuropsychological and neuroimaging studies demonstrate that many cortical (parietal, premotor and prefrontal cortices) and subcortical (basal ganglia and cerebellum) brain structures are involved in the processing of temporal information. However, the functional contribution of these different regions, as well as their interactions, is still subject of considerable debate. Clock-counter models, the most influential cognitive models of temporal computation assign the origin of temporal information to a single mechanism. This hypothetical mechanism can be accessed through various sensory modalities, and is independent from the length of the temporal interval and from the context in which temporal information is used.

In my talk I will present a series of TMS and fMRI studies in which I have challenged the idea of a single centralized and amodal temporal mechanism; demonstrating the existence of modality specific as well as supramodal temporal mechanisms and emphasizing the role of modality specific cortices in temporal computations. In particular I will show that: A) associative areas, like prefrontal and posterior parietal cortices, are important for timing of both auditory and visual stimuli. B) that primary as well as secondary visual and auditory cortices represent the temporal properties respectively of the visual and of the auditory environment.




22 October 2010

1) Yves Wiaux, EPFL, UniGE Methods "Accelerated MRI from compressive sampling"
2) Dimitri VanDeVille, EPFL, UniGE Methods "Surfing the Brain: Wavelet-Based Statistical Parametric Mapping"



Yves Wiaux: The acceleration of the acquisition process, or equivalently the enhancement of the image resolution, is currently an important issue in many MRI applications, ranging from diffusion or functional MRI, to dynamic imaging. Parallel imaging is a now a standard approach in that regard. We will discuss a new, radically different but still complementary approach consisting in pre-modulating the images by quadratic phase profiles before acquisition of incomplete frequency information. This technique stands in sharp contrast with the standard acquisition process consisting in the acquisition of complete frequency information. We will develop a basic intuition underpinning this acquisition strategy in the context of a recent evolution in signal processing theory called ``compressive sampling''. We will also highlight its effectiveness through results of simulations on synthetic data, as well as results of first phantom and in vivo acquisitions at the CIBM 7T scanner at EPFL.

Dimitri VanDeVille: Traditional fMRI data analysis looks for the presence of a hypothetical task-related blood oxygen-level dependent (BOLD) response. If evidence is found (on statistical grounds) voxels are declared as "active". The most popular framework is based on the general linear model (GLM) containing regressors of interest and other covariates. However, the huge number of univariate tests requires proper correction for multiple comparisons. The most widely deployed method uses spatial Gaussian prefiltering to obtain sufficient smoothness such that continuous Gaussian random field theory and Euler characteristics can be applied to yield a lower statistical threshold than the one from (overconservative) Bonferroni correction.

I will explain how the (spatial) wavelet transform can circumvent the need of spatial smoothing while still guaranteeing strong control of false positives without loss of sensitivity compared to Gaussian smoothing. The core of the framework relies on a theorem that bounds the null hypothesis rejection probability after reconstruction from thresholded wavelet coefficients. I will illustrate the pros and cons when applying WSPM to experimental data.

WSPM toolbox:




15 October 2010

1) Artur Marchewka, CHUV Project "Classification of Alzheimer patients using a "rule-out diagnostic" based approach"
2)Holger Sperdin, CHUV Science "Mapping body surface and the spatial co-localization of sounds in auditory-somatosensory multisensory interactions"



Artur Marchewka: There is growing interest in using MRI structural images for robust and reliable classification of Alzheimer's disease (AD) in combination with machine learning tools such as Support Vector Machine (SVM). SVM can achieve high accuracy (85%- 95%) in distinguishing AD patients from healthy controls, however the classification accuracy against other types of dementia decreases dramatically. Another problem arises due to the fact that in routine clinical practice one follows a "rule-out" protocol by excluding alternative diagnosis rather than performing binary classification as in current methods for automated classification. In order to tackle this problems we propose using a one-class classifier for automated "rule-out" diagnosis- based approach. The overall aim is to distinguish the target class (i.e Alzheimer's disease) from all other possible cases (either normal or abnormal), which are considered as outliers. For training, one-class classification uses only cases from the target class and computes a hypersphere that contains this target class. We also explore the impact of diffeomorphic spatial registration (DARTEL/spm8) on the classification results.

Holger Sperdin: Mapping body surface and the spatial co-localization of sounds in auditory-somatosensory multisensory interactions Recently, we have shown that early low-level auditory-somatosensory (AS) multisensory interactions are behaviorally relevant [1-2]. In a subsequent study we addressed the possibility that top-down and/or task-related influences can dynamically impact the spatial representations mediating AS interactions as well as the extent to which multisensory facilitation would be observed or not. Across two psychophysical experiments we show that facilitated detection occurs even when attending to spatial information. Moreover, discrimination performance with probes, quantified using sensitivity (d'), was impaired following multisensory trials in general and significantly more so following misaligned multisensory trials. This indicates that spatial information is not available, despite being task-relevant. The results support a model wherein early AS interactions may result in a loss of spatial acuity for unisensory information [3]. Finally in an ongoing experiment (n=10), we sought to determine whether AS multisensory interactions exhibit sensitivity to body surface and auditory spatial information. Participants detected AS stimuli presented separately or simultaneously while recording 64-channel event-related potentials (ERPs). Either the neck or left hand (situated near the neck) was stimulated. Sounds emanated from nearby or far from the left ear. Reaction times (RTs) were significantly facilitated in excess of probability summation for all multisensory conditions. RTs were faster following stimulation of the hand vs. neck and for nearby vs. far sounds. ERPs were generally stronger when sounds were nearby, irrespective of body surface stimulated, but differences between spatially co-localized and misaligned multisensory conditions occurred 10ms earlier when the neck was stimulated compared to the hand. We discuss these findings in terms of multisensory mapping of body surface and spatial information.

  • 1.Sperdin HF, Cappe C, Foxe JJ, Murray MM. (2009). Early, low-level auditory-somatosensory multisensory interactions impact reaction time speed. Front Integr Neurosci. 2009;3:2.
  • 2.Sperdin HF, Cappe C, Foxe JJ, Murray MM. (2010). The behavioral relevance of multisensory neural response interactions Front. Neurosci., doi: 10.3389/neuro.01.009.2010
  • 3.Sperdin HF, Cappe C, Murray MM. (2010) Auditory-somatosensory multisensory interactions in humans: dissociating detection and spatial discrimination. Neuropsychologia. 2010 Sep 9.




8 October 2010

1) Nathalie Charriere, CHUV Project "Predictive value of structural MRI for stroke recovery"
2) Maria Kynazeva, CHUV Science "Psychogenic seizures and frontal disconnection: EEG synchronization study"



Nathalie Charriere: Stroke is one of the major causes for mortality and severe disability worldwide and millions of people are faced with its consequences. Although stroke-related disabilities have been the subject of a large number of studies, the impact of lesion extent and anatomical location on clinical outcome still remains controversial. The project I will present aims at relating brain lesion topography and extent in order to explain a particular dysfunction. In this purpose, we will analyse the retrospective MRI data of 132 stroke patients. In the long-term, we plan to create a generative model of disease for prediction of clinical outcome in motor, cognitive and emotional/motivational domains. We plan also to evaluate the model for prediction accuracy, regarding the modulation impact of rehabilitation treatment. In parallel, we will analyse the potential benefit of using an MP2RAGE acquisition method instead of MPRAGE. This pilot study will be based on prospective data from 20 acute stroke patients and will help us to chose the proper method to use.

Maria Kynazeva: Psychogenic non-epileptic seizures (PNES) are paroxysmal events that, in contrast to epileptic seizures, are related to psychological causes without the presence of epileptiform EEG changes. Recent models suggest a multi-factorial basis for PNES. A potentially paramount is the interplay between psychiatric features and a specific vulnerability of the brain leading to a clinical picture that resembles epilepsy. Hypothesizing that functional cerebral network abnormalities may predispose to the clinical phenotype, we undertook a characterization of the functional connectivity in PNES patients. We analyzed the whole-head surface topography of multivariate phase synchronization (MPS) in interictal high-density EEG of PNES patients as compared to age- and sex-matched controls. We found widespread inverse correlations between individual PNES frequency and MPS within the prefrontal and parietal cortices. Therefore, prefrontal and parietal hypo-synchronization may reflect subliminal dysfunction of these associative areas in PNES.




1 October 2010

1) Aurelie Manuel, CHUV  Science "Brain dynamics underlying training-induced improvement in suppressing inappropriate action"
2) Sandra DaCosta, EPFL  Science/Methods  "Functional Investigation of the human amygdala at 7 Tesla"



Aurelie Manuel: Inhibitory control, a core component of executive functions, refers to our ability to suppress intended or ongoing cognitive or motor processes. Mostly based on Go/NoGo paradigms, a considerable amount of literature reports that inhibitory control of responses to "NoGo" stimuli is mediated by top-down mechanisms manifesting ~200ms post-stimulus onset within fronto-parietal networks. However, whether inhibitory functions can be trained and the supporting neurophysiological mechanisms remain unresolved. We addressed these issues by contrasting auditory evoked potentials (AEPs) to left-lateralized "Go" and right "NoGo" stimuli recorded at the beginning vs. the end of 30 minutes of active auditory spatial Go/NoGo training (Experiment 1), as well as during passive listening of the same stimuli before vs. after the training session (Experiment 2), generating a 2*2 within-subject design for each experiment. Training improved Go/NoGo proficiency. Response times to Go stimuli decreased. In Experiment 1, AEPs to NoGo, but not Go, stimuli modulated topographically with training 61-104ms post-stimulus onset, indicative of changes in the underlying brain network. Source estimations revealed that this modulation followed from decreased activity within left parietal cortices, which in turn predicted the extent of behavioral improvement. In Experiment 2, by contrast, effects were limited to topographic modulations of AEPs in response to Go stimuli over the 31-81ms interval, mediated by decreased right anterior temporo-parietal activity. We discuss our results in terms of the development of an automatic, bottom-up, form of inhibitory control with training and a differential effect of Go/NoGo training during active executive control vs. passive listening conditions.

Sandra DaCosta: The amydala is thought to play a critical role in human emotional and social behavior, including the recognition of fearful facial expressions.  This presentation will discuss an fMRI study of fearful face processing in the amygdala, conducted at 7T at the EPFL.  The results demonstrate a novel "duration effect" in the amygdala's response to fearful faces with direct vs. averted gaze.  In addition to the scientific questions the technical challenges of scanning the amygdala at 7T will be discussed.




24 September 2010

1) Marzia De Lucia, CHUV Science "A temporal hierarchy for conspecific vocalization discrimination in humans"
2) Ferath Kherif, CHUV Science/Methods "Multivariate Mixture Models and applications to neuroimaging data: lesion detection and characterization of functional individual variability in stroke patients"



Marzia De Lucia The ability to discriminate conspecific vocalizations is observed across species and early during development. However, its neurophysiologic mechanism remains controversial, particularly regarding whether it involves specialized processes with dedicated neural machinery. We identified spatiotemporal brain mechanisms for conspecific vocalization discrimination in humans by applying electrical neuroimaging analyses to auditory evoked potentials (AEPs) in response to acoustically and psychophysically controlled nonverbal human and animal vocalizations as well as sounds of man-made objects. AEP strength modulations in the absence of topographic modulations are suggestive of statistically indistinguishable brain networks. First, responses were significantly stronger, but topographically indistinguishable to human versus animal vocalizations starting at 169-219 ms after stimulus onset and within regions of the right superior temporal sulcus and superior temporal gyrus. This effect correlated with another AEP strength modulation occurring at 291-357 ms that was localized within the left inferior prefrontal and precentral gyri. Temporally segregated and spatially distributed stages of vocalization discrimination are thus functionally coupled and demonstrate how conventional views of functional specialization must incorporate network dynamics. Second, vocalization discrimination is not subject to facilitated processing in time, but instead lags more general categorization by approximately 100 ms, indicative of hierarchical processing during object discrimination. Third, although differences between human and animal vocalizations persisted when analyses were performed at a single-object level or extended to include additional (man-made) sound categories, at no latency were responses to human vocalizations stronger than those to all other categories. Vocalization discrimination transpires at times synchronous with that of face discrimination but is not functionally specialized.

Ferath Kherif  Mixture models are flexible tools for characterizing complex distributions; they are widely used in different domains (from data clustering to pattern recognition). In the neuroimaging field they represent the state of the art for brain segmentation. In this talk, I will present how two apparently different problems can be solved using a Gaussian mixture model (GMM) approach. The first application is the automatic detection of brain lesions in MRI images. The second application is the characterization of the individual variability in structural or functional data. For both cases GMM provides an elegant solution and identifies the most prominent source of intersubject variability to explain differential patterns of structural/functional changes.




17 September 2010

1) Bogdan Draganski, CHUV Science/Methods " Cortico-subcortical connectivity - behind the pretty pictures"
2) Leila Cammoun, EPFL Science/Methods " Mapping the human connectome at multiple scales with diffusion spectrum MRI"



Bogdan Draganski Novel methods based on probabilistic diffusion tractography have been developed in order to capture in detail the topographic organisation of cortico-subcortical projections. I will discuss the strenghts and limitations of these techniques to prove at anatomical level the spatial localisation and extent of nodes within predefined functional networks. Further, I will focus on the potential impact of a reliable framework on studying disease related changes in limbic, associative and sensorimotor circuits.

Leila Cammoun The global structural connectivity of the brain, the human connectome is now accessible to MRI at millimeter scale. We describe a multi-scale approach to map the connectome by constructing normalized whole-brain structural connection matrices derived from diffusion spectrum MRI tractography at 5 different scales. Using a template-based approach to match cortical landmarks of different subjects, we propose a robust method that allows :

  • the selection of identical cortical Regions Of Interest and location in different subjects with identification of the associated fiber tracts
  • straightforward construction and interpretation of anatomically organized whole-brain connection matrices
  • statistical inter-subject comparison of brain connectivity at various scales.

Some current applications of this method will be shortly presented.




10 September 2010     

1) Roberto Martuzzi, EPFL Methods/Science "Finger somatotopy in primary somatosensory cortex in single human subjects: a 7T fMRI study"
2) Silvio Ionta, EPFL Science "A robotics/fMRI joint approach to study subjectivity"



Roberto Martuzzi The primary somatosensory cortex (SI) contains Brodmann areas (BA) 1, 2, 3a, and 3b and electrophysiological recordings in non-human primates showed that BAs 3b, 1, and 2 contain separate representations for each finger. In the present study, we used spatial resolution and BOLD sensitivity available at 7 Tesla to non-invasively investigate the anatomical location of human primary somatosensory cortex. For all subjects we were able to show non-invasively that human SI contains at least three representations of contralateral fingers, where adjacent fingers were represented at adjacent cortical regions across the three SI regions. The spatial organization of SI corresponds well with previous electrophysiological data in non-human primates, but also suggests enlargement of the SI representations of D1.

Silvio Ionta Self-location is abnormal in neurological patients with out-of body experiences (OBEs), it can be studied experimentally in healthy subjects, but its neural underpinnings are still under investigation. Addressing to this lacking information we used visuo-tactile multisensory conflict, MR-compatible robotics, and fMRI. Behavioural results showed that self-location is affected by the synchrony between the tactile stroking applied to the participants' back and the visually presented stroking, only when participants watched a human body (but not an empty room) being stroked. The BOLD signal change in right and left temporo-parietal junction reflected changes in self-location and perspective. Statistical lesion overlap in OBEs patients (using nonparametric voxel based lesion symptom mapping) revealed the involvement of the right temporo-parietal junction, centered at the angular gyrus. The present fMRI and lesion analysis data reveal that the temporo-parietal cortex, especially in the right hemisphere, encodes the experience of the conscious "I" as embodied and localized within bodily space.




3 September 2010

1) Jose Marques, EPFL Methods "7T imaging in a scanner nearby, the pros and challenges"


Jose Marques The talk will cover some of the developments in imaging methodology and results obtained exploring the capabilities of the CIBM human 7T scanner.




16 July 2010

1) Delpine Ribes, EPFL Methods "MRF segmentation and its applications on T1 weighted Brain MR segmentation"
2) Gunnar Krueger, EPFL Methods/Science "High-field MRI physics and techniques"



Delpine Ribes Incorporation of spatial encoding information into an Expectation-maximisation (EM) scheme has been widely used for tissue classification purpose through Markov Random Field (MRF) theory. Several MRF numerical schemes have been proposed for unsupervised image segmentation based on image intensity. I will present three model used to solve numerically EM-MRF for MR. Then based on MR segmentation examples, I will show differences of segmentation results regarding the numerical scheme used.

Gunnar Krueger The presentations will cover high-field specifics (3T and 7T) to be considered for imaging at the 3T and 7T CIBM scanners. Special focus will be on advanced techniques and contrast mechanisms.



9 July 2010

1) Merixtell Bach i Cuadra, EPFL Methods/Science "Challenges of MR tissue segmentation of the fetal brain"
2) Melissa Saenz, CHUV Methods "Tonotopic mapping of human auditory cortex with fMRI at 7T"


Merixtell Bach i Cuadra In this talk I will first present the challenges of fetal brain segmentation in MR images. The ultimately goal of this project is a quantitative analysis of the fetal cortical surface as a marker of the cerebral maturation (as gyration) and also for studying central nervous system pathologies. However, this quantitative approach is a major challenge for several reasons. First, movement of the fetus inside the amniotic cavity requires very fast MRI sequences to minimize motion artifacts, resulting in a poor spatial resolution and/or lower SNR. Second, due to the ongoing myelination and cortical maturation, the appearance of the developing brain differs very much from the homogenous tissue types found in adults. Third, due to low resolution, fetal MR images considerably suffer of partial volume (PV) effect, sometimes in large areas. Today few studies exist related to the automated segmentation of MR fetal imaging. In the second part of the talk I will present a brain tissue classification technique based on an Expectation-Maximization Markov Random Field (EM-MRF) framework and the basal ganglia segmentation based on the Active Contour framework. This work is a joint collaboration with Dr. M. Schaer and Prof. S. Eliez from University of Geneva, School of Medicine, with Dr. L. Guibaud from Hôpital Debrousse, Lyon, and with Prof. J.-Ph. Thiran from EPFL.

Melissa Saenz The auditory cortex in humans and other mammals contains spatial maps of frequency representation. Human tonotopic maps can be identified with functional MRI, allowing for the identification of multiple auditory fields including the primary field (A1). A high-field strength scanner, such as the CIBM 7T is an asset for tonotopic mapping since relatively small functional voxels are needed. I will show human tonotopic maps aquired with the CIBM 7T with an explanation of stimulus selection, data acquisition, analysis and intrepretation of results. Clear tonotopic maps can be acquired with only 8 minutes of scan time.



2 July 2010

1) Micah Murray, CHUV Methods "Challenges in using ERPs in multisensory research"
2) Celine Cappe, CHUV Science "Distinct brain networks and sub-additivity mediate early auditory-visual multisensory interactions in humans"


Micah Murray This methods presentation discusses the issue of how to identify and interpret non-linear multisensory interactions in humans using event-related potentials (ERPs). While there are clear metrics available in single-unit recordings of animals, which have in turn led to the formulation of several principles of multisensory interactions, it is not evident if/how these metrics are transferable to population-level responses in humans. Moreover, canonical ERP analyses pose several interpretational caveats that can be surmounted with recent developments in signal analysis. These methods are applicable to not only multisensory research questions but also more generally.

Celine Cappe Current models of brain organization include multisensory interactions at early processing stages and within low-level, including primary, cortices. Embracing this model with regard to auditory-visual (AV) interactions in humans remains problematic. Controversy surrounds the application of an additive model to the analysis of event-related potentials (ERPs), and conventional ERP analysis methods have yielded discordant latencies of effects and permitted limited neurophysiologic interpretability. While hemodynamic imaging and transcranial magnetic stimulation studies provide general support for the above model, the precise timing, super-/sub-additive directionality, topographic stability, and sources remain unresolved. We recorded ERPs in humans to attended, but task-irrelevant stimuli that did not require an overt motor response, thereby circumventing paradigmatic caveats. We applied novel ERP signal analysis methods to provide details concerning the likely bases of AV interactions. First, non-linear interactions occur at 60-95ms post-stimulus and are the consequence of topographic, rather than pure strength, modulations in the ERP. AV stimuli engage distinct configurations of intracranial generators, rather than simply modulating the amplitude of unisensory responses. Second, source estimations (and statistical analyses thereof) identified primary visual, primary auditory, and posterior superior temporal regions as mediating these effects. Finally, scalar values of current densities in all of these regions exhibited functionally-coupled, sub-additive non-linear effects; a pattern increasingly consistent with the mounting evidence in non-human primates. In these ways, we demonstrate how neurophysiologic bases of multisensory interactions can be non-invasively identified in humans allowing for a synthesis across imaging methods on the one hand and species on the other.




25 June 2010

1) Meritxell Bach i Cuadra, EPFL Methods "Statistical classification methods for brain tissue segmentation in MR images"
2) Dimitri Van de Ville, EPFL Methods/ Science "Inter-subject brain decoding using functional connectivity measures"



Meritxell Bach i Cuadra I will briefly present an overview of brain tissue segmentation techniques in MR images and I will describe the statistical classification in a Bayesian framework including local spatial priors using Markov Random Field theory. The talk will present as well the challenges in the brain tissue segmentation for fetal brain.

Dimitri Van de Ville Recent advances in brain decoding have shown how functional and structural imaging can be exploited advantageously as compared to univariate approaches. In particular, fMRI-based brain decoding relies on BOLD intensity to build a subject-specific classifier, usually in a specific brain region such as visual cortex. In this work, we propose the use of functional connectivity as a marker for brain states that can be deployed for inter-subject decoding. We demonstrate the feasibility of the approach in a movie-watching versus resting-state using leave-one-subject-out crossvalidation. This is joint work with Jonas Richiardi, Hamdi Eryilmaz, Sophie Schwartz, Patrik Vuilleumier.



18 June 2010

1) Bogdan Draganski, CHUV   Welcome and Introduction
2) Gijs Plomp, EPFL Methods/ Science "Inter-subject brain decoding using functional connectivity measures"


Gijs Plomp The visual cortex exhibits functional specialization that can be routinely demonstrated using hemodynamic measures like fMRI and PET. To understand the dynamic nature of cortical processes, however, source imaging with a high temporal resolution is necessary. Here, we asked how well distributed EEG source localization (LAURA) identifies functionally specialized visual processes. We tested three stimulus paradigms commonly used in fMRI with the aim to localize striate cortex, motion-sensitive areas, and face-sensitive areas. EEG source localization showed initial activations in striate and extra-striate areas at around 70 ms after stimulus onset. These were quickly followed by extensive cortical, as well as subcortical activation. Functional motion and face-selective areas were localized with margins of below 2 cm, at around 170 and 150 ms respectively. The results furthermore show for the first time that the C1 component has generators in the insula and frontal eye fields, but also in subcortical areas like the parahippocampus and the thalamus.

Follow us:    
CHUV - Rue du Bugnon 46 - CH-1011 Lausanne
Tel. +41 21 314 95 93
Fax +41 21 314 12 56
Centre Hospitalier Universitaire Vaudois (CHUV)