Brain-Computer Interfaces

Scheme of a Brain-Computer Interface

A Brain-Computer-Interface (BCI) is a device which enables the user to interact with its surrounding only by thought. The GRAZ-BCI is based on the detection of changes in electroencephalogram (EEG) rhythms that are modulated by motor  imagery (MI). MI can be described as mental rehearsal of motor tasks without their execution. Motor tasks could be the imagination of squeezing a training ball, water paddling with both feet or playing an instrument. These thoughts can be used to generate control signal for any device, e.g. a neuroprosthesis or a communication device. 

From the technical point of view, a BCI system operates in 4 phases: 

  1. Signal Acquisition: Brain signals are recorded on the scalp of the users using electrodes, which are mounted on an EEG cap. This happens non-invasive, no harm is done to the user. 
  2. Signal Preprocessing: The measured signals are quite weak, even eye-blinks greatly influence them. Therefore, complex algorithms are applied to enhance the signal quality to reveal the brain patterns
  3. Decoding/Encoding: Preprocessed signals are analyzed with modern machine learn methods to identify brain patterns of the designated imaginations
  4. Control and Feedback: Every action should cause a adequate reaction. If you grasp a glass, you do feel the glass beyond your fingers, get a measure of its weight and feel the temperature of it. This "feedback" helps us performing our tasks of daily life without even fully recognizing them - like adjusting the grasp force we put on the glass when we "feel" that it is heavier than expected. For a person with no sensation in the hand, this sensations cannot be felt anymore. Therefore substitutes have to be implemented - which are called feedback. 

A BCI system works as a closed loop system. Every action the user performs results in any form of feedback to user. An imagined hand movement for instance, could resolve in a command which triggers the movement of a (neuro)prosthesis. The user, allthough he may not be able to feel the movement due to spinal cord injury (SCI), sees the (neuro)prosthesis move his arm. This visual feedback closes the loop. 

BCI use is a skill that users must learn. Although first basis control could be established within a couple of training sessions, usually up to one in four control attempts might be negative. Various studies suggest that intensive BCI training enable users to overcome this limitation. Training BCI skills is often fatiguing and monotonous. Usually users have to repeat different motor imageries, which will later on used to train modern machine learning algorithms and eventually generate control signals out of the individual brain patterns. During this monotonous repetitions, the user often does not get any feedback in how well he performs these actions. Recent studies showed that it is possible to change this by individual adaptation of the system already during the training.

Within the Moregrasp project, BCI will be one of the control modalities for the users. BCI training will happen right away from the start and is an essential part of the individual training for each user. We are aiming to exploit recent adaptive approaches to make training far more interesting than monotonous repetitions of different movements - and we strongly believe that this effort will have positive effects on our users too!

Motor Imagery based Brain-Computer Interfaces

Daniel Hackhofer (Assistent) and Professor Reinhold "Reini" Scherer introduce you to the topic!
(Video is in German)