LINC

CLINICAL APPLICATION OF LINC FOR THE CONTROL OF EXTERNAL DEVICES BASED ON IMPLANTED ELECTRODES


Relevant for Research Area

C - Applications


Summary

Brain-Tune/LiNC harnesses cognitive control signals for manipulation of technical devices via brain activity. Since previous BCI research indicates that information from low-level motor signals alone is too limited for practical applications, we use brain activity from several modalities (movement-, error-, and speech-related tasks) to overcome this problem. Very-high-quality non-invasive and intracranial EEG data from healthy subjects and neurological patients are used to extract discriminative brain activity patterns with cutting-edge deep learning. Deep learning approaches can outperform conventional methods in our experience. However, they are time-consuming to train and often require adjustments of network architectures and multiple hyperparameters to achieve good results. Our goal is to make deep learning on brain data easier applicable and more efficient. All closed-loop systems developed in BrainLinks-BrainTools can profit from improved decoding from neuronal population signals. Brain-Tune/LiNC is thus important to achieve the long-term perspectives of BrainLinks-BrainTools.


Research Status

Brain-Tune/LiNC has made substantial progress in decoding and understanding control signals for brain-computer interfaces (BCIs) including neurorobotic devices. For example, we have recently performed the most comprehensive analysis yet of decodability of different hand movement parameters from a range of intracranial EEG signal components. We showed that movement speed (i.e., the absolute value of velocity) is the overall best-decodable parameter. Furthermore, we came up with a simple explanatory model linking the observed predominance of movement speed over other movement parameters, linking these properties of intracranial EEG as an example of a neural population signal to the well-known properties of motor-cortical single neurons. This novel framework bridges the gap between spatial observation scales of brain activity and aids identification of BCI control signals that are most informative about behavioral states.

Hammer J, Pistohl T, Fischer J, Kršek P, Tomášek M, Marusič P, Schulze-Bonhage A, Aertsen A, Ball T (2016) Predominance of movement speed over direction in neuronal population signals of motor cortex: intracranial EEG data and a simple explanatory model. Cereb Cortex, pp. 1-19.