EEG-based BCI Systems in Neuropsychiatric Diseases

179

the sensors acquire the brain activity, the amplifiers enlarge the amplitude

of EEG signals and the A/D converter digitalizes the signal to store it in a

recording device.

Most of the companies that produce hardware devices for EEG acquisition

also develop software packages to analyze signals, obtain outputs, and pro-

vide feedback [26]. The software developed by a company is mostly commercial

software (e.g., Brain Products, Emotiv). However, there are open-source soft-

ware packages (e.g., EEGlab [27], FieldTrip [28]) that can be run, shared, and

modified by anyone.

7.3

Phases of EEG-based BCI Systems

EEG-based BCI systems are implemented in several sequential steps, starting

with the acquisition of EEG signals followed by pre-processing and feature

extraction/selection phases. After these steps, various artificial intelligence

(AI) techniques fed by the extracted features can be applied to finalize the

BCI system (please see the details of each step in further sub-sections).

7.3.1

Acquisition of EEG Signals

All types of BCI systems require the acquisition of brain signals so this step is

mandatory. To capture and record the desired brain activities, various hard-

ware technologies (please see Section 7.2.4) and well-designed paradigms are

used during the signal-acquisition phase (please see the following sub-section

7.3.2). A recent comprehensive review by Mridha et al. [29] provides a compi-

lation of datasets that encompass BCI paradigms, including motor imagery,

P300-based BCI speller, and cognitive paradigms.

7.3.2

Encoding Paradigms for EEG-based BCI

The paradigms are used during the EEG signal acquisition phase to modu-

late and encode the neural signals via specially designed mental tasks, thus

target brain activities are generated. Since the invention of BCI, multiple

paradigms have been designed to elicit different brain activities such as motor-

related (e.g., MI), sensory-related (e.g., Steady-state visual evoked potential

(SSVEP)), cognitive-related (e.g., Visual and/or Auditory oddball, Go/nogo

paradigms) or Hybrid [30] that corresponds to two or more different mental

strategies (such as EEG signal upon application of visual attention and mo-

tor imagery paradigms). Among various kinds of paradigms, MI, P300 as a

component of event-related potentials and SSVEP are the most widely used

ones in BCI applications. Regarding active BCI, MI paradigm has a common

usage that requires imagining specific movement whereas P300 and SSVEP