Effects of spatial sampling on network alterations in idiopathic generalized epilepsy – can routine EEG be enough?
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Objective
Idiopathic generalized epilepsy (IGE) is characterized by marked brain network alterations as assessed using electrophysiology. The clinical application of high-density EEG or MEG is often hindered by logistical challenges and the need for a volumetric MRI. This study investigates how EEG channel density and the head model influence brain metrics in individuals with IGE versus controls ranging from 256-channel high-density EEG to 19-channel routine EEG.
Methods
Resting-state EEG data from 35 individuals with IGE and 54 healthy controls were collected using a 256-channel setup. Data were analyzed at full density and then iteratively down-sampled to lower densities. Source reconstruction was performed either using individual MRI data or a standard brain template. We assessed EEG power and connectivity group differences at all channel compositions, head model types, and parcellations (cortical vertices, anatomical and network parcellations). Additionally, a clinical sample recorded with 19 channels was analyzed to validate findings in a real epilepsy monitoring scenario (71 patients, 43 controls).
Results
Lower-density arrays reliably identified global group differences for both power and connectivity and in frequency bands for which the strongest effects were observed. The spatial similarity of the results for the 256 channels set and those with less channels were good to moderate for power (r spin ~0.97 to 0.33), but dropped for connectivity with less than 64 channels (r spin ~0.78 to −0.12). Comparing individual and canonical head models revealed consistent effects (r spin ~0.77 to 0.5), with coarser brain parcellations increasing stability for low-density maps.
Significance
Low-density EEG arrays suffice for detecting global alterations in IGE, particularly in signal power. For precision-critical contexts and complex metrics such as connectivity, high-density setups are beneficial. Canonical head models are a viable alternative if no individual MRI is available, especially for regional-or network-level assessments.
Highlights
Averaged EEG power and connectivity alterations in IGE are detectable with low-density EEG
High-density EEG improves spatial accuracy of connectivity estimates
Individual and canonical head models produce comparable group effects on EEG metrics, especially when using anatomical and network parcellations
Our findings advocate for leveraging clinical EEG for network analyses in IGE while emphasizing the need for high-density coverage if spatial precision is needed
