Independent component analysis reveals dynamic ictal BOLD responses in EEG-fMRI data from focal epilepsy patients

Abstract

Introduction-Seizures occur rarely during EEG-fMRI acquisitions of epilepsy patients, but can potentially offer a better estimation of the epileptogenic zone than interictal activity. Independent component analysis (ICA) is a data-driven method that imposes minimal constraints on the hemodynamic response function (HRF). In particular, the investigation of HRFs with clear peaks, but varying latency, may be used to differentiate the ictal focus from propagated activity.

Imaging haemodynamic changes related to seizures: Comparison of EEG-based general linear model, independent component analysis of fMRI and intracranial EEG

“…Our findings of ICs spatially concordant with the seizure onset zone, complement a recent study in which spatially concordant ICs were identified in patients in whom a GLM analysis had previously revealed seizure-related changes (Levan et al, 2010) . We have extended the ICA approach by applying it to all cases regardless of the GLM approach and used the gold standard of icEEG for validation rather than the HRF.…”

“…We expected to find "BOLD" components with a spatial pattern involving sub-cortical structures, (Federico et al, 2005;Levan et al, 2010;Tyvaert et al, 2008), but none were observed. This may be because the classifier was trained for the cortex and sub-cortical components are generally smaller in size, so the degree of clustering on which ICA depends may not be comparable (Formisano, 2009).…”

“…In this study we considered the temporal characteristics of interictal and ictal events and therefore included components classified as EPI artefacts, to account for potential low-frequency ictal BOLD signal changes (Federico et al, 2005;Tyvaert et al, 2008) and inspected all ICs spatially concordant with the seizure onset zone. This method differs from the approach of fitting an HRF to component time courses (Levan et al, 2010) .…”

“…Early ictal fMRI studies without EEG, found an increase in BOLD signal at or in advance of seizure onset compared to baseline, within the presumed seizure onset zone (Detre et al, 1995) (Federico et al, 2005). Mathematically rigorous signal processing methods such as Independent Component Analysis (ICA) applied to simulated and interictal data, have demonstrated BOLD changes concordant with the results of traditional EEG-derived fMRI modelling techniques Rodionov et al, 2007) opening a new avenue for the investigation of seizurerelated fMRI changes (Levan et al, 2010).…”

Resting-state fMRI studies in epilepsy

“…Having both fairly high spatial and temporal resolu- analysis [68][69][70] , graph theory [61,62] , and dynamic causal modeling [64] . Applications of these methods to epilepsy need close collaboration among multi-disciplinary investigators, especially when multiple techniques (e.g., structural MRI, diffusion MRI, EEG, as well as fMRI) are required.…”