Modeling of the BOLD signal using event-related simultaneous EEG-fMRI and convolutional sparse coding analysis

Area of Science:

• Neuroscience
• Cognitive Science
• Biomedical Engineering

Background:

• Simultaneous electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) offer complementary insights into brain activity.
• Extracting reliable neural event markers from EEG is crucial for accurate fMRI analysis.
• Transient events in EEG are challenging to isolate and utilize effectively.

Purpose of the Study:

• To evaluate convolutional sparse coding (CSC) for transient event detection in EEG.
• To compare EEG-derived activation maps with fMRI data and response time (RT) measures.
• To assess the utility of CSC-detected events for hemodynamic response function (HRF) estimation in fMRI.

Main Methods:

• Acquired simultaneous EEG-fMRI data during a visually-guided attention task.
• Applied convolutional sparse coding (CSC) to extract transient EEG events.
• Performed voxel-wise fMRI analysis using CSC events and compared with RT-based analysis.
• Estimated HRF shapes using FIR models with CSC-detected events.

Main Results:

• Demonstrated concordance between fMRI activation maps derived from CSC events and RT-based analysis.
• Observed consistent hemodynamic response function (HRF) shapes across subjects using CSC events.
• Validated CSC as a viable tool for identifying reliable neural events in EEG.

Conclusions:

• Convolutional sparse coding (CSC) is a promising technique for extracting meaningful transient events from EEG.
• The integration of CSC with simultaneous EEG-fMRI enhances the precision of brain activity analysis.
• This approach offers a robust method for event detection and HRF estimation in neuroscience studies.