Lemieux – ACES / Europe Dublin 2015 Performing electrophysiological measurements in humans inside Magnetic Resonance Imaging scanners; applications in Epilepsy research and other areas Louis Lemieux Department of Clinical and Experimental Epilepsy UCL Institute of Neurology, Queen Square, London & MRI Unit, Epilepsy Society Chalfont St Peter, Buckinghamshire UK
43
Embed
Performing electrophysiological measurements in humans inside Magnetic Resonance Imaging scanners; applications in Epilepsy research and other areas
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Performing electrophysiological measurements in humans inside Magnetic
Resonance Imaging scanners; applications in Epilepsy research and other areas
Louis Lemieux
Department of Clinical and Experimental EpilepsyUCL Institute of Neurology, Queen Square, London
&MRI Unit, Epilepsy Society
Chalfont St Peter, BuckinghamshireUK
Lemieux – ACES / Europe Dublin 2015
Outline
• Epilepsy
• Multimodal neuroimaging in humans
– EEG and functional MRI (fMRI)
– Mapping epileptic events using scalp EEG-fMRI
• Going deeper: intracranial EEG-fMRI
– Technique implementation
– An epileptic seizure
• Conclusions
Lemieux – ACES / Europe Dublin 2015
Rely on accurate anatomical targeting
Epilepsy• Epilepsy is the most common serious chronic neurological
condition affecting all ages– 50 million people affected in the world– Economic costs (~€20 billion/year for Europe).
• 30% of all people with epilepsy have seizures that do not respond to medical treatment, leading to:– Cognitive decline– Poor quality of life– Significantly increased mortality– High societal costs
• Need for improved treatment– Surgery– Drug delivery
• Need for improved localisation of the epileptogenic areas
Lemieux – ACES / Europe Dublin 2015
• fMRI
– Allows tomographic visualisation of haemodynamic changes associated with
brain activity
– Has better temporal resolution than PET (…for epileptic spikes)
– Is non-invasive (BOLD)
• EEG
– Important observable of brain activity in humans
– Reflects neuronal signal generation and synchronisation
– Important clinical tool in epilepsy (epileptic spikes, seizures, etc)
– Non-invasive (scalp) & cheap
Basic principles: EEG & fMRI
Lemieux – ACES / Europe Dublin 2015
LF= 0.5 Hz HF= 30 Hz
1 sec.
25 uV
10:36:33 10:36:34 10:36:35 10:36:36
Fp2-F8
F8-T4
T4-T6
T6-O2
Fp1-F7
F7-T3
T3-T5
T5-O1
ECG1-ECG2
Epileptic spikes
• Spatially linked to the epileptic focus
• Brief (<100ms), unpredictable
• Sub-clinical: can only be seen on EEG
• Epileptic spike source localisation– Inverse problem of EEG…
– Can fMRI better localise epileptic spike generators?
– Clinical utility?
Lemieux – ACES / Europe Dublin 2015
EEG-fMRI in epilepsyData acquisition strategy
Subject at restSimultaneous EEG-MRI
• fMRI:− Echo-planar imaging (EPI) fMRI
scanning sequences
− Whole-brain coverage
• EEG:− 64-channel cap on scalp
− MR-compatible amplifier and digitiser
− Digital signal transmitted to recording laptop outside the scanner room
Louis Lemieux
Amps placed at far end of scanner bore.
Lemieux – ACES / Europe Dublin 2015
EEG-fMRI of epileptic spikes
Continuous EEG and fMRI:
EEG-based GLM
Spike-related BOLD
[Krakow et al, 1999; Lemieux et al, 2001]
1 sec.
25 uV
10:36:33 10:36:34 10:36:35 10:36:36
Fp2-F8
F8-T4
T4-T6
T6-O2
Fp1-F7
F7-T3
T3-T5
T5-O1
ECG1-ECG2
Lemieux – ACES / Europe Dublin 2015
Fp1-Pz
F7 – Pz
T3 – Pz
T5 – Pz
O1 – Pz
Fp2 – Pz
F8 – Pz
T4 – Pz
T6 – Pz
O2 – Pz
Fp1 – F7
F7 – T3
T3 – T5
T5 – O1
Fp2 – F8
F8 – T4
T4 – T6
T6 – O2
ECG
OSC
1 Second LF = 0.5 Hz HF = 45 Hz50uV
The intra-MRI EEG artefact problem
Lemieux – ACES / Europe Dublin 2015
Fp1-Pz
F7 – Pz
T3 – Pz
T5 – Pz
O1 – Pz
Fp2 – Pz
F8 – Pz
T4 – Pz
T6 – Pz
O2 – Pz
Fp1 – F7
F7 – T3
T3 – T5
T5 – O1
Fp2 – F8
F8 – T4
T4 – T6
T6 – O2
ECG
OSC
1 Second LF = 0.5 Hz HF = 45 Hz50uV
[Allen et al, 2000]
The intra-MRI EEG artefact problem
Lemieux – ACES / Europe Dublin 2015
Fp1-Pz
F7 – Pz
T3 – Pz
T5 – Pz
O1 – Pz
Fp2 – Pz
F8 – Pz
T4 – Pz
T6 – Pz
O2 – Pz
Fp1 – F7
F7 – T3
T3 – T5
T5 – O1
Fp2 – F8
F8 – T4
T4 – T6
T6 – O2
ECG
OSC
1 Second LF = 0.5 Hz HF = 45 Hz50uV
[Allen et al, 1999]
The intra-MRI EEG artefact problem
Lemieux – ACES / Europe Dublin 2015
Clinical relevance: EEG-fMRI of epileptic spikes inFocal Cortical Dysplasia
ANN NEUROL 2011;70:822–837
Conclusion:Scalp EEG-fMRI of epileptic spikes may predict less promising surgical cases and therefore avoid unnecessary invasive interventions
Lemieux – ACES / Europe Dublin 2015
icEEG: Subdural grids and depth electrodes• Surgically placed on cortex• Sampling similar to high-density scalp EEG• Used to map epileptogenic tissue in relation to eloquent
cortex• 6x8 array of 3mm diameter Pt-Ir disk contacts
[Fried et al, 1999]
• Surgically inserted within brain• Used to detect epileptogenicity and propagation in deep
cortex/lesions• Sensitivity profile very different from scalp EEG and grids:
‘tunnel vision’ [see Cosandier et al 2007; Church et al, 1985]
• Health hazards• RF-induced heating• Induced voltages (stimulation)• Factors considered:• Field strength: 1.5T and 3T• RF transmit coil type: head and body• Electrodes: depths and grids • EEG wires
• Geometry and placement• Length• Termination
Simultaneous icEEG-fMRI:Safety tests in phantoms
Lemieux – ACES / Europe Dublin 2015
[Carmichael et al, 2009, 2012]
• RF-induced heating greatest health risk• Excessive heating observed• Body coil• 3T• But for certain realistic conditions:
Max ∆T = 0.9ºC (@1.5T, SAR=2.4W/Kg, 6mins)
icEEG-fMRI possible without excessive additional health risk under certain conditions
Site-specific assessment necessary
Intra-cranial EEG-fMRI safety study results:Heating tests
Lemieux – ACES / Europe Dublin 2015
1) Use likely safest regime2) Strict protocol:
- 1.5T Siemens Avanto / head (quad.) Tx/Rx coil- 90cm cables with 10cm fold along scanner central Z xis- Foam insert designed for exact positioning
- Position EEG system and cables reproducibly- Low SAR sequences:
- T1 volume, gradient echo EPI [TE=40ms], B0 map- 3-4% of 3.2 W/Kg- max duration 10 minutes
3) Close monitoring and documenting of patient responses, images, appearance of brain surface, histology
icEEG-fMRI: Implementation
Lemieux – ACES / Europe Dublin 2015
icEEG-fMRI image quality:EPI signal degradation around electrodes
[Carmichael et al, 2012]
Lemieux – ACES / Europe Dublin 2015
icEEG-fMRI image quality:EPI signal degradation around electrodes
[Carmichael et al, 2012]
Lemieux – ACES / Europe Dublin 2015
Feasibility study case report:• Seizures: R hand stiffening• MRI normal• No spikes during scalp EEG-fMRI• icEEG implantation: grids and strips over
left frontal lobe• Two 10-minute resting-state icEEG-fMRI
sessions:• 100’s of L fronto-central spikes
L
icEEG-fMRI
MEG: IED onset
L
L
MEG: IED propagation
L
L
irritative zone
Demonstration of spike-correlated icEEG-fMRI
[Vulliemoz/Carmichael et al, 2010]
Lemieux – ACES / Europe Dublin 2015
Conclusions
EEG-fMRI allows:• Haemodynamic mapping of events with specific EEG features• ‘Extends’ EEG: whole-brain coverage, source complexity-independent
Intracranial EEG-fMRI:• Fascinating, complex data• Exquisite electrophysiological sensitivity• Image data quality is an issue:
− Electrode composition is suboptimal for MR imaging− Clinical impact: Limits ability to locate the electrodes in relation to the
anatomy− Research impact: Reduces the amount of fMRI signal available for
analysis
Lemieux – ACES / Europe Dublin 2015
Team / collaborators / funding
David W. CarmichaelUmair ChaudharyAna-Carolina Coan (Campinas)Alessio De Ciantis (Firenze)Beate DiehlJohn S. DuncanMadeline GradeMarco LeiteAndrew McEvoyTeresa MurtaIrene PappalardoSuejen Perani (King’s)Sofia Markoula (Ioannina)Roman RodionovCatherine ScottNiraj SharmaRachel C. ThorntonAndré van GraanAnna Vaudano (Modena)Matthew C. WalkerBritta Wandschneider+ The Clinical Neurophysiology and
Neuroradiology teams at the National Hospital for Neurology and Neurosurgery, Queen Square
K Friston (UCL)M Guye / F Bartolomei / P Chauvel / JP Ranjeva (Marseille)S Vulliemoz / C Michel (Geneva)P Figueiredo (Lisbon)M Papadopoulou / D Marinazzo (Ghent)Radhakrishnan A / Chandrasekharan K / Sreedharan S (Trivandrum)R Quian Quiroga / C Pedreira (Leicester)V Kokkinos (Thessaloniki)S Meletti (Modena)F Cendes (Campinas)K Mullinger / R Bowtell (Nottingham)KR Muller / V Samek / D Blythe (Berlin)H Laufs (Frankfurt)J Daunizeau (Paris)K Whittingstall (Sherbrooke)E Formisano / F De Martino (Maastricht)M Torkmani Azar (Konya)
Funding:Action Medical ResearchBrain Research Trust / James Tudor FoundationNIHR (UK Department of Health)Medical Research CouncilSupport in kind:Brain Products
Thank you
Lemieux – ACES / Europe Dublin 2015
The problem of presurgical test validation• Validation of localisation tests in epilepsy is fundamentally limited• Gold standard is not golden:
• Surgical resection localisation and outcome data
• Ictiogenic ‘source’ must be considered a network in most patients a priori
• Elements of a possible solution• Improved characterisation / Modelling
– deep sleep, learning, motivational processes and reward system• Theta (4-7 Hz):
– working memory, emotional arousal and fear conditioning• Alpha (8-12 Hz):
– cortical operations during the awake resting-state in the absence of sensory inputs, disengagement of task-irrelevant brain areas, working memory and short-term-memory retention
SPM{F} contrast across all IZ1 IED-related effects
Summary measure for comparison:Relationship of BOLD clusters with the presumed, icEEG-derived EZ
Lemieux – ACES / Europe Dublin 2015
Conclusions (2)
• EEG-fMRI (+video) to have increased importance in neuroscience• Intracranial EEG-fMRI
• Analysis is complex: additional layer on top of icEEG analysis & interpretation• Abundance of activity -> more reliable maps than scalp EEG-fMRI• Interpretation: compare it against what?
• How precious is our ‘gold’ standard?• Surgical outcome + localisation of resected tissue: what about disruption of
wider networks?
• Effective connectivity: ‘the full multimodal generator model’ (DCM)• Improved biophysical models of epileptogenic networks being developed• Applicable to icEEG, then fMRI
Lemieux – ACES / Europe Dublin 2015
Seizure propagation in hypothalamic hamartomas (HH)
• Seizures originate in HH, and control of seizures can be achieved by surgically removing the HH.
• Possible surgical alternative: disconnection of the underlying pathway.
• Different seizure propagation pathways have been described in HH [Leal et al. Epilepsia 2003; Kahane et al. Epileptic Disord. 2003]:
• Aim: identify the correct seizure propagation pathway in individual patients using DCM.
1.HH to temporal-occipital (posterior, PR) to frontal lobe (anterior, AR)
(through the fornix)
2.HH to frontal (anterior, AR) to temporal-occipital (posterior, PR) lobe
(through the mammillo-thalamo-cingulate pathway)
[Murta et al, 2012]
Lemieux – ACES / Europe Dublin 2015
Seizure propagation in HH:Family of DCM models
• HH is the driving region• All structures consistent with each of 2 propagation hypotheses• Each connection: uni- or bi-directional (22=4 structures per hypothesis)• Each connection: linear (black) or bilinear (black and green) (24=8 models per hypothesis)
[Murta et al, 2012]
Lemieux – ACES / Europe Dublin 2015
Seizure propagation in HH: Bayesian model comparison
[Murta et al, 2012]
Most likely model
Model 1:
HH → temp-occ. → frontal
Beyond mapping:Networks and causality(and a detour via IGE)
Lemieux – ACES / Europe Dublin 2015
• Multi-modal imaging = Combinations of images or maps– from different sources (instruments)
or– that show different aspects (e.g. MR contrasts)
• Fundamental assumption: measurements relate to the same phenomenon– Location– Time
Multi-modal imaging: basics
Lemieux – ACES / Europe Dublin 2015
Estimating a network of sources and their dynamics:Dynamic Causal Modelling
[Friston, 2009; David et al 2006]
Bi-linear model of effective connectivity
– Effect of activity in one region on activity in other: equations of motion
– Biophysical generative model
fMRI:EEG/MEG neural mass model:
+ Model comparison (Bayesian)
Lemieux – ACES / Europe Dublin 2015
EEG fMRI of GSW - QS 1.5T seriesGroup analysis
IGE
N=18
Sup Post Par ↓
Front ↓
Post Cing ↓
Thalam ↑
[Hamandi et al., 2006]
Th
Th
Lemieux – ACES / Europe Dublin 2015
Dynamic causal models ofeffective connectivity in GSW
Centrencephalic
Cortico/corticoreticular
Precuneus model
[Vaudano et al, 2009]
Data:EEG-fMRI of GSW in 7 patients with IGESignificant GSW-related BOLD: Thalamus, Precuneus and ventromedial prefrontal
Model of effective connectivity: DCM of fMRIGSW EEG onsets and offsets modelled as (endogenous) input3 models:
Best in 5/7 cases
Lemieux – ACES / Europe Dublin 2015
Integrate and Fire Neurons
Mean Field Population Dynamics
DCM of focal seizuresA more realistic biophysical model of (fast) ictal activity
Time (s)
Freq
uenc
y (H
z)
5 10 15 20 25 30
8
16
32
64
128
Model: Coupled neural population dynamics
[M Leite, unpublished]
Lemieux – ACES / Europe Dublin 2015
+10mV
-70mV
E vs. I
Pyr Int
Cur
rent
(pA
)
Phase (%) Phase (%)
Phase (%) Phase (%)
Cur
rent
(pA
)C
urre
nt (
pA)
Current (pA) Current (pA)
Pyr Int
U
[M Leite; unpublished - In collaboration with D Kullmann and D Kuzmin]
A more realistic biophysical modelof (fast) ictal activity: Validation initial results