Figure 1.
Effective connectivity models.
Effective connectivity (DCM) models showing GSW discharges as autonomous input on three different regions (dotted arrows) within the cortical thalamic system: 3 ROI are structurally (forward and backward) connected (solid arrows). Model A: GSW as autonomous input on the thalamus; Model B: GSW as autonomous input on the ventromedial prefrontal cortex (vmPFC). Model C: GSW as autonomous input on the precuneus (BA 7). GSW: Generalised Spike and Wave discharges; BA: Brodmann Area.
Table 1.
Clinical details of patients studied based on ILAE diagnostic categories.
Table 2.
DCM Regions of interest.
Figure 2.
Representative example of EEGs recorded during scanning after scanning artefact subtraction.
The EEG traces were analysed following pulse (not shown) and image artifact subtraction; EEG traces are displayed as bipolar montage. OSC: scanner slice pulse used for EEG artifact correction, and EEG-fMRI synchronization (7/s). (A) IGE/JAE: patient (ID #7). The trace shows an epoch of 3.5 Hz generalised spike-wave complexes (length ∼4 seconds) with anterior predominance. (B): IGE/JME: patient (ID #21a).The trace shows an epoch of 2.5−3 Hz generalised multispike-wave complexes (length ∼2.5 seconds) with anterior predominance.
Figure 3.
EEG-FMRI statistical parametric map in a patient with JME.
A colour-coded overlay of SPM{t} (red: positive BOLD response; green: negative BOLD response) (p<0.05 corrected for Family-Wise Error-FWE) onto the slices overlay shows, BOLD signal increase in bilateral cingulated gyrus (BA32) and BOLD signal decrease in bilateral thalamic, bilateral caudate, right medial frontal gyrus (BA10), left superior temporal gyrus (BA39), right precuneus (BA7), bilateral inferior parietal lobuli (BA39). Clusters labelling according to Talairach Daemon, (http://ric.uthscsa.edu/project/talairachdaemon.html). BA: Brodmann Area
Table 3.
EEG-fMRI results.
Figure 4.
Effective connectivity model comparison results.
Bayesian Model Selection (BMC) among DCMs for the three models tested. (a): differences between log-evidences for model A, B, C for each subject. The triangles identify the best model on the basis of the subject's highest log-evidence difference. A difference greater than 3 is highly significant. For illustration purposes we added a constant value of 100 to all log-evidence differences. The numbers inside brackets (on the x-axis) correspond to ID no in Hamandi et al. [23]. (b): graph showing the difference between the log-evidence at the group level, i.e. pooled over subjects, for the three models. For illustration purposes we added a constant value of 100 to all log-evidence differences.
Table 4.
DCM F values.