Mapping brain activity using event-related independent components analysis (eICA): specific advantages for EEG-fMRI

Area of Science:

• Neuroimaging
• Epilepsy Research
• Signal Processing

Background:

• Standard functional MRI (fMRI) analyses assume canonical hemodynamic responses, which may not apply to all neural events, such as interictal epileptiform discharges (IEDs).
• Non-canonical hemodynamic responses (HRF) and pre-event activity in conditions like epilepsy can complicate fMRI interpretation.
• Less constrained analyses are needed but risk conflating neural signals with artifacts like motion or physiological noise.

Purpose of the Study:

• To develop and validate an event-related independent components analysis (eICA) method for fMRI.
• To differentiate neural activity from artifacts in event-related fMRI data.
• To objectively identify neural components consistent across subjects in group analyses.

Main Methods:

• Developed an event-related independent components analysis (eICA) framework.
• Implemented a group analysis to identify spatially and temporally consistent eICA components across subjects.
• Applied eICA to fMRI data from patients with rolandic epilepsy exhibiting IEDs.

Main Results:

• The eICA method successfully identified distinct sources of event-related signal changes.
• The approach allowed for the assessment and separation of potential artifacts from neural signals.
• In rolandic epilepsy patients, a single eICA component, localized to the expected source of IEDs, was consistently detected.

Conclusions:

• eICA offers a robust method for analyzing event-related fMRI data, especially when canonical assumptions are violated.
• This technique enhances the ability to detect and interpret neural activity, such as IEDs, by separating them from artifacts.
• The findings demonstrate eICA's utility in identifying neural sources in specific patient populations like those with rolandic epilepsy.