skip to primary navigation skip to content

CBSU bibliography search

To request a reprint of a CBSU publication, please click here to send us an email (reprints may not be available for all publications)

Group analysis of distributed-dipole solutions: Statistical parametric mapping (SPM) applied to word-pseudoword differences in MEG
International Conference on Biomagnetism, Sapporo, Japan, 25/8 - 29/8 2008
Year of publication:
CBU number:
Many approaches exist for estimating the neural sources of MEG signals; however, few of these methods allow for group-level analysis and population inference. Statistical parametric mapping (SPM) may be applied to distributed-dipole source models to allow for such inference. The present study used this approach to indentify the brain regions underlying visual word recognition. Subjects (15 right-handed adult English speakers) silently read words and pronounceable pseudowords (480 each; order randomised; duration 300 ms) presented sequentially at fixation and indicated by button-press whether each was a word or nonsense word (response-hand mapping counterbalanced). MEG data were acquired (1000 Hz) using a 306-sensor whole-head array (102 magnetometers, 204 planar gradiometers; Elekta Neuromag). Neuromag’s MaxFilter utility was used to remove external and sensor-space noise (signal-space separation with spatio-temporal extention) and to compensate for head movements. Data were downsampled to 333 Hz and imported into Matlab for analysis in SPM and EEGLAB. Blink artefacts were removed using independent component analysis; epochs (-100ms to 1000 ms) containing residual artefact or noise were excluded. Grand average global field power of the word-nonword difference revealed two peaks between 300-500 and 500-700 ms, respectively. For each subject, distributed-dipole source localisation was performed (several approaches were evaluated) on epochs, to include induced and evoked energy, using dipoles constrained to the subject’s cortical surface. The resulting 3D image volumes were smoothed (12-mm gaussian kernel) and normalised to permit group analysis. Group SPMs revealed a network of brain regions consistent with findings from fMRI: clusters of ‘active’ voxels (words > pseudowords) were found in bilateral superior and inferior temporal gyrus; clusters showing the opposite pattern were found in left anterior and inferior frontal regions, premotor cortex, and precuneus.