Related Experiment Video
Updated: Jul 3, 2026
11:28
Concurrent EEG and Functional MRI Recording and Integration Analysis for Dynamic Cortical Activity Imaging
Published on: June 30, 2018
Sparse linear regression for reconstructing muscle activity from human cortical fMRI
1Department of Computational Neurobiology, ATR International, Computational Neuroscience Laboratories, 2-2-2 Hikaridai, Keihanna Science City, Seika-cho, Soraku-gun, Kyoto, 619-0288, Japan. gganesh@atr.jp
Neuroimage
|July 19, 2008
Summary
This study demonstrates that functional magnetic resonance imaging (fMRI) can reconstruct individual muscle activity. Our novel Bayesian approach decodes neural signals for non-invasive motor control research.
Area of Science:
- Neuroscience
- Motor Control
- Brain Imaging
Background:
- Invasive methods are not feasible for mapping individual muscle activity to human brain signals.
- Current non-invasive brain imaging lacks the spatial resolution to isolate specific muscle-related neural activity.
Purpose of the Study:
- To investigate the feasibility of reconstructing individual muscle activity from functional magnetic resonance imaging (fMRI) data.
- To develop and validate a novel algorithm for decoding muscle activity using fMRI signals.
Main Methods:
- Simultaneous recording of surface electromyography (EMG) and fMRI during isometric wrist muscle tasks.
- Application of Bayesian sparse regression to map fMRI activity in motor cortices (M1, pre-motor, SMA) to EMG.
- Validation of the developed mapping on an independent dataset, comparing it with support vector machine and least square regression.
Main Results:
- Successfully reconstructed individual muscle activity from fMRI data.
- The Bayesian sparse regression model demonstrated superior generalization compared to conventional decoding algorithms.
- Identified intermingled yet distinct voxel sets in M1 and pre-motor cortex associated with antagonist muscle activity, extending beyond traditionally recognized wrist control regions.
Conclusions:
- fMRI data, analyzed with Bayesian linear models, can predict individual human muscle activity.
- The developed algorithm offers a novel, non-invasive tool for studying neural mechanisms of motor control and learning.
