Temporal representation of arm force direction using fNIRS signals

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Functional near-infrared spectroscopy (fNIRS) is a non-invasive neuroimaging technique.
• Decoding brain activity from fNIRS signals is an active area of research.
• Understanding motor control relies on accurately interpreting neural signals.

Purpose of the Study:

• To investigate the feasibility of creating a temporal representation of brain activity from fNIRS signals.
• To determine if fNIRS data can be used to estimate the direction of isometric arm movements.
• To compare the effectiveness of sparse linear regression (SLR) and support vector machine (SVM) for decoding fNIRS data.

Main Methods:

• Subjects performed isometric arm movements in four directions.
• fNIRS signals were recorded over the left primary motor cortex.
• Sparse linear regression (SLR) and support vector machine (SVM) classifiers were employed to estimate arm force direction.
• Classification accuracy was evaluated for both methods.

Main Results:

• Classification accuracy for arm force direction reached approximately 70% using SLR.
• The temporal patterns of features identified by SLR and SVM were comparable.
• fNIRS signals demonstrated potential to contain information about arm force direction within 4-6 seconds post-stimulus.

Conclusions:

• fNIRS signals captured information related to arm force direction.
• Machine learning classifiers, particularly SLR, can decode motor intentions from fNIRS data.
• These findings suggest the potential of fNIRS for real-time monitoring of motor control.