Interictal epileptic networks in MEG and SEEG · 2013-11-14 · Interictal epileptic networks in MEG and SEEG Christian Bénar Institut de Neurosciences des Systèmes, Marseille

Institut de Neurosciences des Systèmes

Interictal epileptic networks in MEG and SEEG

Christian Bénar Institut de Neurosciences des Systèmes, Marseille

[email protected]

Workshop on Scale-free Dynamics and Networks in

Neurosciences, October 21-24, 2013, Montréal

http://www.inserm.fr/


Overview • Context

• Characterization of networks

– Spike versus oscillations – Strategy 1: Source localization – Strategy 2: ICA

• Simultaneous recordings of EEG/MEG/SEEG







Presurgical Evaluation • The goal of presurgical evaluation is to remove

the epileptogenic zone « the minimum amount of cortex to be removed in order to render the

patient seizure-free » H. Lüders « the zone of primary organization of seizures » Talairach and Bancaud

MEG/EEG SEEG

(intracerebral EEG)

Semiology

Phase 1 Phase 2 Surgery


Magnetoencephalography

Marseille MEG platform, 248 sensors

1 2 3

4 5 6

L R

EEG

L R 1 2 3

4 5 6

MEG

Gavaret et al Brain Topogr in press


Stereotaxic EEG (SEEG)

Barbeau et al 2008: evoked activity during face recognition

SEEG implantation in deep regions


Objectives

MEG/EEG SEEG (intracerebral EEG)

Phase 1 Phase 2 Surgery

Optimize signal processing strategies by confronting MEEG with SEEG

Guide SEEG implantation

Avoid SEEG in some cases


Seizure start: focus or network ?

Bartolomei et al 2001 Start of seizure mesial + neocortical

Bartolomei et al 2008 Epileptogenicity index


What about interictal activity? • Non-invasive tools mostly investigate interictal

discharges (i.e., between seizures)

• Simpler propagation patterns than seizures

• Complex relationship with seizure (Gotman 1989)

• Present network dynamic patterns: indication on propagation pathways?


Existence of extended interictal networks in SEEG


« preictal build up »

Schwartz et al 2011

The interictal networks fluctuate in time


Gavaret et al J Clin Neurophysiol 2006

time (s)

frequ

ency

(Hz)

2 2.05 2.1 2.15 2.2 2.25 2.3 2.35 2.4 2.45 2.510

15

20

25

30

35

40

45

50

55

60

MEG discharge at 30 Hz

Interictal spikes and oscillations Average spike

Interictal activity can manifest as spikes, oscilation or a mixture of both (« moustache ») Transient patterns (not 1/f !!)







Rationale • Characterize interictal epileptic networks, leading

region(s) with non-invasive tools

• Investigate differences betwen spikes and oscillations

• Strategies

– Time-frequency /time-scale analysis – Source localization – Independent component analysis (ICA)




– Spike vs oscillations: methodological points – Strategy 1: Source localization – Strategy 2: ICA



Point 1: frequency overlap

-100

0

100

Original Signalam

plitu

de (u

V)

-50

0

50Filtered Signal (8-12 Hz)

time (s)

freq

uenc

y (H

z)

Time-Frequency Analysis

200 400 600 800 1000 1200 1400248

153058

Transients and oscillations overlap in frequency: difficulty of extracting « pure » oscillatory part


« False ripples » (1) : transients

Bénar et al 2010

Filtering spikes or artefacts gives « false » oscillations


« False ripples » (2): harmonics

Filtering signals with harmonics gives also « false » oscillations

Bénar et al 2010


Separation of oscillations and transients Oscillations

Jmail et al 2010

In real data, dependence on fit between atoms and data Current collaboration with S Hitziger, T Papadopoulo, M Clerc (INRIA Sophia Antipolis): definition of atoms from data


Point 2: multi-trial aspects

Benar et al 2009

« consensus matching pursuit » -> Taking into account interevent variability in atomic decomposition


Point 3: Interactions background/activity

Indication of phase resetting in primary visual cortex

Krieg et al 2010







Spike networks vs oscillation networks

N. Jmail


Spike networks vs oscillation networks

Nawel Jmail

-non-invasive results concordant with SEEG: 89 % for spikes , 75 % for oscillations). -Only partial overlap between networks (38 % of total number of regions in MEG, 50% in SEEG). -Current directions: automatize extraction of networks, increase number of patients







ICA on continuous MEG

Malinowska et al, Hum Brain Mapp 2013


Co-occurrence testing

Similar principle than Bourien et al 2005

Institut de neurosciences des systèmes

Co-occurrence graphs

• significant direction

• spike co-occurrences (significance with non-parametric technique)


• leaders: delays or conditional probability

Source localisation on components

ICA1

ICA7

ICA6

ICA1

ICA6

ICA7


Resultats (1): good concordance

L

MEG

SEEG



L

MEG

SEEG

L

Results (2): partial concordance



Comparison with ictal networks Interictal, SEEG Ictal, SEEG



Networks: discussion • One can reliably extract interictal networks from

non-invasive MEG data

• One must be cautious on possible undetectable leaders: need to investigate depth/surface relationships

• Need to validate clinically on larger population of patients (seizure freedom)

• Future direction: use of virtual brain simulations







Rationale

• Previous comparison MEEG/SEEG done on separate recordings

• Not guaranteed to visualize exact same activity (temporal fluctuations)

• Cannot use single trial fluctuations as source of information (see simultaneous EEG-fMRI)

• Need to record simultaneous data


Recording and Artefacts


Simultaneous MEG-EEG-SEEG

Dubarry et al, under review


Source localization



Single trial analysis



Simultaneous recordings: discussion

• We have shown the feasibility of simultaneous EEG/MEG/intracerebral EEG - Average - Single trial analysis

• Issue of artefacts: need further work on denoising

• Future directions: mapping of sensitivity of measures according to location/extent of activity

Acknowledgments

Neurologists P Chauvel F Bartolomei M Gavaret A Trébuchon Da Fonseca

Neuropsychology C Liegeois-Chauvel

Engineers JM Badier S Chen B Colombet M Woodman B Giusiano

Students, postdocs U Malinowska N Jmail AS Dubarry

Neurosurgeons J Régis R Caron

Collaborations: T Papadopoulo, M Clerc (INRIA Sophia); X Alario (LNC, Marseille); F Wendling (LTSI, Rennes) M Guye, JP Ranjeva (CEMEREM, Marseille), C Grova, JM Lina (Montreal)

Funding: Agence National de la Recherche (ANR Multimodel) Fondation pour la recherche médical (FRM)

Documents

Interictal epileptic networks in MEG and SEEG · 2013-11-14 · Interictal epileptic networks in MEG and SEEG Christian Bénar Institut de Neurosciences des Systèmes, Marseille