You are here: Home News Publications Hand movement direction detected in infrared light signals from the brain
Document Actions

Hand movement direction detected in infrared light signals from the brain

New recording technique closes gap in approaches towards brain-machine interfaces

Hand movement direction detected in infrared light signals from the brain

The setup to detect the near-infrared light signal (black-and-white image) is rather simple. Different hand movements result in different levels of oxygenation in hemoglobin over time (inset, red and blue solid curves).

The activity of nerve cells in areas of the cortex responsible for body movements is tuned to different parameters, such as hand movement direction. These signals can be detected and used in order to create brain-machine interfaces (BMIs), which allow the direct control of a computer or a robotic arm through brain activity. Neuronal activity creates a number of different signals that can be exploited. The most accurate is the pulse-like activity of a single nerve cell, but this is only detectable with sensors in the brain. Furthermore, electrical activity of thousands and millions of nerve cells in any area of the brain sum up and give rise to neuronal population activity. A tuning of such signals in relation to movement directions has been found not only for spiking activity but also for neuronal population activity (which can be measured in different forms: local field potentials, electrocorticography, electroencephalography, magnetoencephalography, or functional magnetic resonance imaging).

A non-invasive and indirect technique to record neuronal population activity is the so-called functional near-infrared spectroscopy (fNIRS) that measures the intensity of near-infrared light that has travelled through the cortex. This intensity depends on the strength of light absorption by oxygenated and deoxygenated hemoglobin, whose concentrations in the blood in turn are influenced by the neuronal activity (a phenomenon called the BOLD effect). Recently, researchers working on motor control and BMIs have become interested in this technique.
As fNIRS is portable (an advantage over magnetic resonance imaging, which requires a huge recording apparatus) and not corrupted by electromagnetic noise (an advantage over portable electroencephalographic devices), it might prove useful for BMIs. However, so far it had been unclear whether fNIRS signals are actually tuned to the direction of hand movements – and which performance can be expected in corresponding BMIs.

In a new study, published in the scientific journal PLoS ONE, researchers from Freiburg and London recorded fNIRS signals at several brain areas while the participants in this experiment moved one hand in one of two directions. In addition, a magnetic tracking system recorded the participants’ head movements in order to assess their influence on the recorded data. The direction of movements of one hand could be decoded with an average accuracy of 65 % from fNIRS signals in the opposite brain hemisphere. This accuracy is highly significant but rather small – determining movements purely by chance would already yield an accuracy of 50 %. Here, it is important to note that decoding movement parameters from unilateral hand movements is strikingly different from decoding left hand versus right hand movements. The latter BMI approach can exploit that each hand is controlled mainly by the opposite hemisphere.

fNIRS signals from the hemisphere on the same side as the used hand did not contain information about the movement direction. Although the participants were instructed to prevent head movements, the magnetic tracking data revealed that small, involuntary, direction correlated head movements did occur. Due to the high precision of the tracking system, these head movements could be decoded with very high accuracy. Importantly, the researchers could show that such head movements did not affect the decoding of fNIRS signals.

These results demonstrate for the first time that also fNIRS signals vary with the direction of hand movements, though only weakly. This finding closes a gap in the spectrum of BMI relevant recording techniques of brain activity. However, as the decoding accuracy is quite low compared to many other recording techniques, the researchers conclude that fNIRS is currently not suitable for practical BMIs that use decoding of movement direction.

Still, fNIRS could play a role in the experimental settings of future research. It proved to be relatively resistant to head movements, so it might be attractive for studies that investigate brain activity during motor experiments.

Original publication:
Stephan Waldert, Laura Tüshaus, Christoph P. Kaller, Ad Aertsen and Carsten Mehring (2012) fNIRS exhibits weak tuning to hand movement direction. PLoS ONE 7(11): e49266. doi:10.1371/journal.pone.0049266

Personal tools