1 Components of a BCI system: signals from the user’s brain are acquired and processed to extract specific features used for classification. The classifier output is transformed into a device command, which, at the same time, provides feedback to the user. 

Context in source publication

Context 1

... brain–computer interface (BCI) transforms signals originating from the human brain into commands that can control devices or applications. In this way, a BCI provides a new nonmuscular communication channel, which can be used to assist patients who have highly compromised motor functions, as is the case with patients suffering from neurological diseases such as amyotrophic lateral sclerosis (ALS) or brainstem stroke. The immediate goal of current research is to provide these users with an opportunity to communicate with their environment. Present-day BCI systems use different electrophysiological signals such as slow cortical potentials, evoked potentials, and oscillatory activity recorded from scalp or subdural electrodes, and cortical neuronal activity recorded from implanted electrodes. Due to advances in methods of signal processing, it is possible that specific features automatically extracted from the electroencephalogram (EEG) and electrocorticogram (ECoG) are used to operate computer-controlled devices. The interaction between the BCI system and the user, in terms of adaptation and learning, is a challenging aspect of any BCI development and application. This chapter outlines and explains current approaches and methods used in BCI research. A technical system that permits direct communication between brain and computer is known as a BCI. 1,2 In this case, the normal communication channels, such as speech and movement, are not used, but instead the brain activity is directly recorded and transformed into a control signal. Therefore, a BCI provides a new communication channel that can be used to convey messages and commands directly from the brain to the external world. The use of a BCI depends on the interaction of two adaptive controllers, the user’s brain and a computer, which has to produce an action that accomplishes the user’s intention. One general base for a BCI is that music or visual or motor imagery modifies neuronal activity and can result in measurable changes in firing patterns of cortical neurons and in the ongoing EEG and ECoG. 3,4 Furthermore, focused or selective attention can enhance different brain signals or components of evoked potentials and in turn can be used in a BCI system. 5 Beside electrical potentials, the blood oxygen level dependent (BOLD) response measured by real-time functional magnetic resonance imaging (fMRI) can be used as an input signal for a BCI. 6 The current and most important applications of a BCI are the restoration of a communication channel for patients with a locked-in syndrome and the control of neuroprosthesis in patients with spinal cord injury. Aside from these, there are the important and established field of neurofeedback therapy and the upcoming field of multimedia and virtual reality applications. In this context, a BCI could be used for diverse tasks such as playing games or providing multidimensional feedback in virtual reality. Concerning the last point, BCI adds a new dimension in man–machine interaction and may be of great importance in multimedia applications. A BCI system is, in general, composed of the following components: signal acquisition, preprocessing, feature extraction, classification (detection), and application interface (Figure 14.1). The signal acquisition component is responsible for recording the electrophysiological signals providing the input to the BCI. Depending on the type of analyzed brain signals and the processing mode, sensory stimulation may be necessary. The task of preprocessing is to enhance the signal-to-noise ratio. This can include artifact reduction methods and the application of advanced signal processing methods. After preprocessing, the signal is subjected to the feature extraction algorithm. The goal of this component is to find a suitable representation (signal features) of the electrophysiological data that simplifies the subsequent classification or detection of brain patterns. That is, the signal features should encode the commands sent by the user, but should not contain noise and other patterns that can impede the classification process. There are a variety of feature extraction methods used in current BCI systems. A list (not exhaustive) of these methods includes amplitude measures, band power, Hjorth parameters, autoregressive models, and wavelets. The task of the classifier component is to use the signal features provided by the feature extractor to assign the recorded samples of the signal to a category of brain patterns. In the simplest form, detection of a single brain pattern is sufficient, for instance, by means of a threshold method. 7,8 More sophisticated classifications of different patterns depend on linear or nonlinear classifiers. 9 The classifier output, which can be a simple on–off signal or a signal that encodes a number of different classes, is transformed into an appropriate signal that can then be used to control a variety of devices. For most current BCI systems, the output device is a computer screen and the output is the selection of certain targets. Advanced applications include the controlling of spelling systems or other external apparatuses such as prosthetic devices and multimedia applications. Many BCI systems, including animal applications, 10 use immediate feedback of performance. In human applications feedback of performance is usually given by visualization of the brain signal on a computer screen or the presentation of an auditory or tactile analogue of the actual brain response (mu rhythm, slow cortical potential, or other EEG activity). The mode of operation determines when the user performs a mental task and intends therewith to transmit a message. In principle, this step can be divided into two distinct modes of operation, the first being externally paced (cue-based, com- puter-driven, synchronous BCI) and the second being internally paced (noncue- based, user-driven, asynchronous BCI). In the case of a synchronous BCI, a fixed, predefined time window is used. After a visual or an auditory cue stimulus, the subject has to act and produce a specific mental state. Nearly all known BCI systems work in such a cue-based mode. 9,11,12 An asynchronous protocol requires a continuous analysis and feature extraction of the recorded brain signal. Thus, such BCIs are, in general, even more demanding and more complex than BCIs operating with a fixed timing scheme. In a synchronous BCI, for example, only two mental tasks or two brain states have to be differentiated, whereas in an asynchronous BCI, a third brain state has to be identified, which is the resting or idling state, also referred to as zero class. To date, only a few BCI research groups are working on an asynchronous BCI. 8,13–18 In principle, there are invasive and noninvasive methods used to record brain signals. The EEG, a noninvasive method, records electrical potential changes and reflects the common activity of several millions of neurons extending over some square centimeters of the cortical tissue. Invasive methods are exemplified by the ECoG as well as by intracortical recordings. In contrast to the EEG, the ECoG represents integrated bioelectrical activity over a much smaller cortical area, but still constitutes the common activity of many thousands of neurons. The multichannel intracortical recordings reflect extracellular activity generated by small neuronal populations in the order of about 100 cells or fewer. 19,20 In the EEG, as well as in the EcoG, two types of phenomena need to be ...