State-space estimation of the input stimulus function using the Kalman filter: a communication system model for fMRI...

Area of Science:

• Neuroimaging
• Signal Processing
• Cognitive Neuroscience

Background:

• The blood oxygenation level-dependent (BOLD) signal in functional magnetic resonance imaging (fMRI) is subject to temporal delays and distortion.
• This blurring is caused by the voxel hemodynamic impulse response function (IRF), affecting the accuracy of brain activity measurements.

Purpose of the Study:

• To reconstruct the input stimulus function from fMRI data by accounting for the IRF.
• To evaluate fMRI experiments as a communication system, treating the stimulus as a message transmitted through a channel characterized by the IRF.
• To gain insight into actual brain activity during task activation with reduced temporal blurring.

Main Methods:

• Utilizing knowledge of the voxel IRF, obtained either during or separately from the fMRI experiment.
• Dynamically estimating the input stimulus function by deconvolution with the IRF.
• Comparing the reconstructed input stimulus function (received message) with the original stimulus (transmitted message) to determine transmission error rates.

Main Results:

• Successful reconstruction of the input stimulus function from fMRI data.
• Demonstration of fMRI as a communication system where the IRF acts as a channel.
• Quantification of transmission errors between the original stimulus and the reconstructed signal.

Conclusions:

• Reconstruction of the input stimulus function provides a more accurate estimate of brain activity during fMRI tasks.
• This method reduces temporal blurring, offering clearer insights into neuronal responses.
• It represents a foundational step towards estimating the true neuronal input function.