Attention Training System Based on EEG Pattern Recognition and Virtual Reality
Tan, Yunchuan (2019-12-27)
Attention Training System Based on EEG Pattern Recognition and Virtual Reality
(27.12.2019)
Julkaisu on tekijänoikeussäännösten alainen. Teosta voi lukea ja tulostaa henkilökohtaista käyttöä varten. Käyttö kaupallisiin tarkoituksiin on kielletty.
suljettu
Julkaisun pysyvä osoite on:
https://urn.fi/URN:NBN:fi-fe202001223050
https://urn.fi/URN:NBN:fi-fe202001223050
Tiivistelmä
The attention level of individual has an important impact on everyone's learning and daily life. Some attention deficit diseases, such as attention deficit hyperactivity disorder (ADHD), are widespread in children to adolescents. Patients with ADHD often have difficulty in maintaining concentration for a long time, behavioral impulsiveness and learning difficulties, which have serious impacts on learning and life. With the rapid development of brain-computer interface technology, virtual reality technology and neural feedback technology, a large number of researchers have begun to combine information science and cognitive neuroscience to conduct research on attention levels and ADHD treatment.
This paper proposes an attention training sytem based on attention level for children with ADHD. There are three key factors in real-time games based on EEG. One is to design attention-sensitive cognitive tasks for racing control; the other is to use EEG acquisition channels as little as possible to improve the portability of the system; third, the attention classification reaches a certain degree of accuracy. This paper studies these three key factors. The main contents of this paper include system design, EEG signal processing and pattern recognition, and the design and development of attention training games.
This paper optimizes the classification model, including data calibration and EEG acquisition channels optimization. After data calibration, the accuracy rate of single-feature model can be improved by 5%. Under the single-feature model, the β-band power and Katz fractal dimensions under the FC6 channel achieved good classification results. The accuracy of the 8-fold cross-validation was 90.9317% and 91.3224%, respectively. With all features of the ten channels, the accuracy can be as high as 97.598%.
At the same time, this paper explores the channels of data acquisition and seeks the optimal data acquisition channel. The four channels of FC6, AF3, F7, and F8 have achieved good results in the classification of attention. Using these four channels can achieve a classification accuracy rate of 97.2187%. On the basis of guarantee accuracy, the number of required channels is greatly reduced, and the portability of the system is improved.
Finally, this paper validated this system in the attention training system and proved that it has a good ability to distinguish between different levels of attention, and proved the feasibility of the system.
This paper proposes an attention training sytem based on attention level for children with ADHD. There are three key factors in real-time games based on EEG. One is to design attention-sensitive cognitive tasks for racing control; the other is to use EEG acquisition channels as little as possible to improve the portability of the system; third, the attention classification reaches a certain degree of accuracy. This paper studies these three key factors. The main contents of this paper include system design, EEG signal processing and pattern recognition, and the design and development of attention training games.
This paper optimizes the classification model, including data calibration and EEG acquisition channels optimization. After data calibration, the accuracy rate of single-feature model can be improved by 5%. Under the single-feature model, the β-band power and Katz fractal dimensions under the FC6 channel achieved good classification results. The accuracy of the 8-fold cross-validation was 90.9317% and 91.3224%, respectively. With all features of the ten channels, the accuracy can be as high as 97.598%.
At the same time, this paper explores the channels of data acquisition and seeks the optimal data acquisition channel. The four channels of FC6, AF3, F7, and F8 have achieved good results in the classification of attention. Using these four channels can achieve a classification accuracy rate of 97.2187%. On the basis of guarantee accuracy, the number of required channels is greatly reduced, and the portability of the system is improved.
Finally, this paper validated this system in the attention training system and proved that it has a good ability to distinguish between different levels of attention, and proved the feasibility of the system.