Section: New Results

Combined fMRI, M/EEG and dMRI

White Matter Information Flow Mapping from Diffusion MRI and EEG.

Participants : Samuel Deslauriers-Gauthier, Jean-Marc Lina [ETS - Ecole de Technologie Supérieure, Montréal] , Russel Butler [Université de Sherbrooke, Sherbrooke] , Kevin Whittingstall [Université de Sherbrooke, Sherbrooke] , Pierre-Michel Bernier [Université de Sherbrooke, Sherbrooke] , Maxime Descoteaux [Université de Sherbrooke, Sherbrooke] , Rachid Deriche.

The human brain can be described as a network of specialized and spatially distributed regions. The activity of individual regions can be estimated using electroencephalography and the structure of the network can be measured using diffusion magnetic resonance imaging. However, the communication between the different cortical regions occurring through the white matter, coined information flow, cannot be observed by either modalities independently. Here, we present a new method to infer information flow in the white matter of the brain from joint diffusion MRI and EEG measurements. This is made possible by the millisecond resolution of EEG which makes the transfer of information from one region to another observable. A subject specific Bayesian network is built which captures the possible interactions between brain regions at different times. This network encodes the connections between brain regions detected using diffusion MRI tractography derived white matter bundles and their associated delays. By injecting the EEG measurements as evidence into this model, we are able to estimate the directed dynamical functional connectivity whose delays are supported by the diffusion MRI derived structural connectivity. We present our results in the form of information flow diagrams that trace transient communication between cortical regions over a functional data window. The performance of our algorithm under different noise levels is assessed using receiver operating characteristic curves on simulated data. In addition, using the well-characterized visual motor network as grounds to test our model, we present the information flow obtained during a reaching task following left or right visual stimuli. These promising results present the transfer of information from the eyes to the primary motor cortex. The information flow obtained using our technique can also be projected back to the anatomy and animated to produce videos of the information path through the white matter, opening a new window into multi-modal dynamic brain connectivity.

This work has been published in [11].

Structural connectivity to reconstruct brain activation and effective connectivity between brain regions

Participants : Brahim Belaoucha, Théodore Papadopoulo.

Understanding how brain regions interact to perform a specific task is very challenging. EEG and MEG are two non-invasive imaging modalities that allow the measurement of brain activation with high temporal resolution. Several works in EEG/MEG source reconstruction show that estimating brain activation can be improved by considering spatio-temporal constraints but only few of them use structural information to do so. In this work, we present a source reconstruction algorithm that uses brain structural connectivity, estimated from diffusion MRI (dMRI), to constrain the EEG/MEG source reconstruction. Contrarily to most source reconstruction methods which reconstruct activation for each time instant, the proposed method estimates an initial reconstruction for the first time instants and a multivariate autoregressive model that explains the data in further time instants. This autoregressive model can be thought as an estimation of the effective connectivity between brain regions. We called this algorithm iterative Source and Dynamics reconstruction (iSDR). This paper presents the overall iSDR approach and how the proposed model is optimized to obtain both brain activation and brain region interactions. The accuracy of our method is demonstrated using synthetic data in which it shows a good capability to reconstruct both activation and connectivity. iSDR is also tested with real data (face recognition task). The results are in phase with other works published with the same data and others that used different imaging modalities with the same task showing that the choice of using an autoregressive model gives relevant results.

This work has been submitted to the non-invasive brain imaging special issue of Journal of Neural Engineering.

Estimation of Axon Conduction Delay, Conduction Speed, and Diameter from Information Flow using Diffusion MRI and MEG.

Participants : Samuel Deslauriers-Gauthier, Rachid Deriche.

The different lengths and conduction velocities of axons connecting cortical regions of the brain yield information transmission delays which are believed to be fundamental to brain dynamics. While early work on axon conduction velocity was based on ex vivo measurements , more recent work makes use of a combination of diffusion Magnetic Resonance Imaging (MRI) tractography and electroencephalography (EEG) to estimate axon conduction velocity in vivo. An essential intermediary step in this later strategy is to estimate the inter hemispheric transfer time (IHTT) using EEG. The IHTT is estimated by measuring the latency between the peaks or by computing the lag to maximum correlation on contra lateral electrodes. These approaches do not take the subjects anatomy into account and, due to the limited number of electrodes used, only partially leverage the information provided by EEG. In our previous work, we proposed a method, named Connectivity Informed Maximum Entropy on the Mean (CIMEM), to estimate information flow in the white matter of the brain. CIMEM is built around a Bayesian network which represents the cortical regions of the brain and their connections, observed using diffusion MRI tractography. This Bayesian network is used to constrain the EEG inverse problem and estimate which white matter connections are used to transfer information between cortical regions. In our previous work, CIMEM was used to infer the information flow in the white matter by assuming a constant conduction velocity for all connections. In this context, the conduction speed, and thus the delays, were inputs used to help constrain the problem. Here, we instead assume that the connection used to transfer information across the hemispheres is known, due the design of the acquisition paradigm, but that its conduction velocity must be estimated.

This work has been published in [23].

Estimation of Axonal Conduction Speed and the Inter Hemispheric Transfer Time using Connectivity Informed Maximum Entropy on the Mean

Participants : Samuel Deslauriers-Gauthier, Rachid Deriche.

The different lengths and conduction velocities of axons connecting cortical regions of the brain yield information transmission delays which are believed to be fundamental to brain dynamics. A critical step in the estimation of axon conduction speed in vivo is the estimation of the inter hemispheric transfer time (IHTT). The IHTT is estimated using electroencephalography (EEG) by measuring the latency between the peaks of specific electrodes or by computing the lag to maximum correlation on contra lateral electrodes. These approaches do not take the subject's anatomy into account and, due to the limited number of electrodes used, only partially leverage the information provided by EEG. Using the previous published Connectivity Informed Maximum Entropy on the Mean (CIMEM) method, we propose a new approach to estimate the IHTT. In CIMEM, a Bayesian network is built using the structural connectivity information between cortical regions. EEG signals are then used as evidence into this network to compute the posterior probability of a connection being active at a particular time. Here, we propose a new quantity which measures how much of the EEG signals are supported by connections, which is maximized when the correct conduction delays are used. Using simulations, we show that CIMEM provides a more accurate estimation of the IHTT compared to the peak latency and lag to maximum correlation methods.

This work has been published in [24].

A Unified Model for Structure–function Mapping Based on Eigenmodes

Participants : Samuel Deslauriers-Gauthier, Rachid Deriche.

Characterizing the connection between brain structure and brain function is essential for understanding how behaviour emerges from the underlying anatomy. To this end, a common representation of the brain is that of a network, where nodes represent cortical and sub–cortical gray matter volumes and edges represent the strength of structural or functional connectivity. A convenient representation of this network is that of a matrix, where entries represent the strength of the structural connectivity (SC) or functional connectivity (FC) between nodes. A number of studies have shown that the network structure of the white matter shapes functional connectivity, leading to the idea that it should be possible to predict the function given the structure. A strategy is to learn a direct mapping from the SC matrix to the FC matrix. In this work, we show that the mappings currently proposed in the literature can be generalized to a single model and that this model can be used to generate new structure-function mappings. We tested our general model on 40 subjects of the Human Connectome Project and demonstrated that for specific choices of parameters, our model reduces to previously proposed models and yields comparable results. However, by allowing to choose the eigenvalue and eigenvector mapping independently, our models can also produce novel mapping that improve the prediction of FC from SC.

This work is currently under submission to OHBM.

Connectivity-informed spatio-temporal MEG source reconstruction: Simulation results using a MAR model

Participants : Ivana Kojcic, Théodore Papadopoulo, Samuel Deslauriers-Gauthier, Rachid Deriche.

Recovering brain activity from M/EEG measurements is an ill-posed problem and prior constraints need to be introduced in order to obtain unique solution. The majority of the methods use spatial and/or temporal constraints, without taking account of long-range connectivity. In this work, we propose a new connectivity-informed spatio-temporal approach to constrain the inverse problem using supplementary information coming from diffusion MRI. We present results based on simulated brain activity using a Multivariate Autoregressive Model, with realistic subject anatomy obtained from Human Connectome Project dataset.

This work has been published in [35].

Connectivity-informed solution for spatio-temporal M/EEG source reconstruction

Participants : Ivana Kojcic, Théodore Papadopoulo, Samuel Deslauriers-Gauthier, Rachid Deriche.

Recovering brain activity from M/EEG measurements is an ill–posed problem and prior constraints need to be introduced in order to obtain unique solution. The majority of the methods use spatial and/or temporal constraints, without taking account of long–range connectivity. In this work, we propose a new connectivity–informed spatio–temporal approach to constrain the inverse problem using supplementary information coming from difusion MRI. We present results based on simulated brain activity obtained with realistic subject anatomy from Human Connectome Project dataset.

This work has been published in [34].

Deconvolution of fMRI Data using a Paradigm Free Iterative Approach based on Partial Differential Equations

Participants : Isa Costantini, Samuel Deslauriers-Gauthier, Rachid Deriche.

Functional magnetic resonance imaging (fMRI) is a technique which indirectly measures neural activations via the blood oxygenated level dependent (BOLD) signal. So far, few approaches have been proposed to regularize the fMRI data, while recovering the underlying activations at the voxel level. In particular, for task fMRI, voxels time courses are fitted on a given experimental paradigm. To avoid the necessity of a priori information on the pattern, supposing the brain works with blocks of constant activation, Farouj et al. has developed a deconvolution approach which solves the optimizations problem by splitting it into two regularization problems, i.e. spatial and temporal. Starting from this idea, we propose a paradigm-free iterative algorithm based on partial differential equations (PDEs) which minimizes the image variations, while preserving sharp transitions (i.e. brain activations), in the space and the time dimensions at once.

This work has been published in [27].

Novel 4-D Algorithm for Functional MRI Image Regularization using Partial Differential Equations

Participants : Isa Costantini, Samuel Deslauriers-Gauthier, Rachid Deriche.

State-of-the-art techniques for denoising functional MRI (fMRI) images consider the problems of spatial and temporal regularization as decoupled tasks. In this work we propose a partial differential equations (PDEs)-based algorithm that acts directly on the 4-D fMRI image. Our approach is based on the idea that large image variations should be preserved as they occur during brain activation, but small variations should be smoothed to remove noise. Starting from this principle, by means of PDEs we were able to smooth the fMRI image with an anisotropic regularization, thus recovering the location of the brain activations in space and their timing and duration.

This work has been published in [28].

Spatially Varying Monte Carlo Sure for the Regularization of Biomedical Images

Participants : Marco Pizzolato [Signal Processing Lab (LTS5), EPFL, Lausanne] , Erick Jorge Canales-Rodríguez [Radiology Department CHUV, Lausanne] , Jean-Philippe Thiran [Signal Processing Lab (LTS5), EPFL, Lausanne] , Rachid Deriche.

Regularization, filtering, and denoising of biomedical images requires the use of appropriate filters and the adoption of efficient regularization criteria. It has been shown that the Stein's Unbiased Risk Estimate (SURE) can be used as a proxy for the mean squared error (MSE), thus giving an effective criterion for choosing the regularization amount as to that minimizing SURE. Often, due to the complexity of the adopted filters and solvers, this proxy must be calculated with a Monte Carlo method. In practical biomedical applications, however, images are affected by spatially-varying noise distributions, which must be taken into account. We propose a modification to the Monte Carlo method, called svSURE, that accounts for the spatial variability of the noise variance, and show that it correctly estimates the MSE in such cases.

This work has been published in [30].

The visual word form area (VWFA) is part of both language and attention circuitry

Participants : Lang Chen, Demian Wasserman, Daniel Abrams, John Kochalka, Guillermo Gallardo-Diez, Vinod Menon.

While predominant models of visual word form area (VWFA) function argue for its specific role in decoding written language, other accounts propose a more general role of VWFA in complex visual processing. However, a comprehensive examination of structural and functional VWFA circuits and their relationship to behavior has been missing. Here, using high-resolution multimodal imaging data from a large Human Connectome Project cohort (N = 313), we demonstrate robust patterns of VWFA connectivity with both canonical language and attentional networks. Brain-behavior relationships revealed a striking pattern of double dissociation: structural connectivity of VWFA with lateral temporal language network predicted language, but not visuo-spatial attention abilities, while VWFA connectivity with dorsal fronto-parietal attention network predicted visuo-spatial attention, but not language abilities. Our findings support a multiplex model of VWFA function characterized by distinct circuits for integrating language and attention, and point to connectivity-constrained cognition as a key principle of human brain organization.

This work has been published in [10].