MfD EEG/MEG Source Localization 4 th Feb 2009 Maro Machizawa Himn Sabir Expert: Vladimir Litvak
Inverse problem 1.Existence 2.Unicity 3.Stability
1.Existence 2.Unicity 3.Stability Inverse problem
1.Existence 2.Unicity 3.Stability Inverse problem Introduction of prior knowledge is needed
Spatio-temporal modeling
Spatio-temporal modeling – step 1 Load EEG/MEG file
Spatio-temporal modeling – step 2 Name the analysis (optional)
Spatio-temporal modeling – step 3 Create/load meshes Bigger the parameter, better the resolution of the results
Spatio-temporal modeling – step 4 Coregister fiducial points with MRI Choose either of methods to coregister –“select” from default locations (at FIL) –“type” MNI coordinates directory –“click” manually each fiducial point from MRI images
Spatio-temporal modeling – step 4 Coregister fiducial points with MRI
Spatio-temporal modeling – step 5 Forward model
Spatio-temporal modeling – step 5 Bayesian model inversion
Spatio-temporal modeling – step 5 Invert: alternative models GS (greedy search: default): –iteratively add constraints (priors) ARD (automatic relevance determination): –iteratively remove irrelevant constraints COH (coherence): –LORETA-like smooth prior IID (independent identically distributed): –minimum norm
Spatio-temporal modeling – step 5 Invert: alternative models The bigger the number, the better the model
Spatio-temporal modeling – step 5 Invert: visualization options 1 digit (ms): map on that time(ms) 2 digits (ms): video during the period 3 digits (x y z): max. voxel on that MNI coordinate
Spatio-temporal modeling – step 6 Window : Induced: localization on each single trial then averaged Evoked: localization on already averaged data INDUCED IMAGE
Spatio-temporal modeling – step 7 Image
Group analysis: same analysis on multiple subjects
(Optional step5) Variational Bayes Equivalent Current Dipole
Optional: time-voltage display