A Comparative Analysis of Ergonomic and Neurophysiological Features of a Hybrid Eye-Brain-Computer Interface and Brain-Computer Interfacesдипломная работа (Магистр)
Организация, в которой проходила защита:
National Research University – Higher School of Economics
Год защиты:2015
Аннотация:Currently, brain-computer interfaces (BCI) are becoming more and more popular in various fields of science and technology. Open new horizons of interaction between people and techniques. Allowing achieve significant successes in medicine, for the rehabilitation of patients with disorders of motor functions, as well as for technical and gaming industry (Ganin, 2013).
In the case of medical applications of BCI, existing keyboards allow patients to make information contact with the environment: by typing of words, giving any commands to the program. Also, to interact with real physical objects, in this case, it refers to the various types of artificial limbs and exoskeletons (Ganin, 2013).
In general, BCI is an advanced technology that allows the subject to interact with a variety of external devices based on the process of recording a brain activity and without using muscles (Allison et al., 2007).
In modern neuroscience of one the preferred method of detecting changes in activity in the brain is EEG (electroencephalography), which is easily accessible and inexpensive method with respect to the other. Of course, it is worth to mention that EEG have own difficulties and constraints, which are discussed below. Along with the aforementioned techniques another is used: MEG (Magnetoencephalography), NIRS (Near-infrared spectroscopy), fMRI (Functional magnetic resonance imaging). The BCI used more then one method of recording data from the users are called hybrid brain-computer interface (hBCI) (Allison et al., 2007).
Many developed BCI have not intuitive system of control of an interface. Observance of a certain schemes, needed to shift electrical brain potentials in the right direction (what would be allowed to work as closely as possible classifier) require training which can lasts for hours or even day (Allison et al., 2007). It is not very comfortable situation for patients which also prevents the formation of automated skills of controlling a BCI. In the case where the interface is designed for paralyzed people, this factor is one of the main one. Researchers must be particularly attentive and sensitive to the needs of future users.
In connection with the above, in the laboratory for neuroeconomics and brain-computer interfaces at the NRC Kurchatov Institute developed the hBCI, with “single-stimulus” paradigm: one target stimulus and the four possible positions up-down, left-right. Such a design of the interface is very easy to operate so that it is possible to reduce training time. Another important factor is the number of mistakes made by the classifier at command recognition. In this case, the existing neurointerfaces also require modifications and improvements. The proposed in this paper design of hBCI based on the well-known P300 BCI design (which discussed in the theories revive in this paper) combined with detection of saccadic eye movements in response to stimuli can be one of the solution for this issues. It is expected that recognition of the user's command based on saccade and on EEG will both benefit from this combination, and, therefore, fast and reliable command recognition will be possible.
This paper presents a theoretical overview of a number of existing non-invasive BCI. The possibility of improving the existing BCI designs is discussed. Some controversial issues are presented: “Midas touch problem”, low accuracy of the classifiers and necessity in lengthy trainings for users to control the interfaces. In the research proposal following the review, plan for testing of developed hBCI is presented and the results and possible directions of future work are discussed.
According to all of the above the aim of this work is to test and optimize BCI based on the P300 combined with detection of saccadic eye movements in response to stimuli.
Based on the aim of research the following research objectives are putting forward:
• Describe the main types of BCI: their characteristics, limitations.
• Identify the potential problems arising from the use of various hBCI and their possible solutions (low accuracy of classifiers, “Midas touch problem”).
• Describe examples of hBCI bases on P300 BCI combined with the detection of saccades appeared in response to the stimulus.
• Consider the possible algorithms for saccades detection.
• Describe the plan of empirical research and to identify the expected results and future directions of the work, as well as possible restrictions.