Analysis of mental imagery based cognitive tasks for brain computer interface
By representing the EEG signals (brain waves) recorded during mental imagery in terms of features and classifying them using an appropriate classifier, the mental imagery tasks performed can be identified accurately and thus be used for BCI in full applications. The optimal electrodes for mental ima...
| Main Author: | |
|---|---|
| Format: | Final Year Project / Dissertation / Thesis |
| Published: |
2019
|
| Subjects: | |
| Online Access: | http://eprints.utar.edu.my/3910/ http://eprints.utar.edu.my/3910/1/fyp_EE_2019_TCH.pdf |
| _version_ | 1848886023274102784 |
|---|---|
| author | Tang, Chee Hoe |
| author_facet | Tang, Chee Hoe |
| author_sort | Tang, Chee Hoe |
| building | UTAR Institutional Repository |
| collection | Online Access |
| description | By representing the EEG signals (brain waves) recorded during mental imagery in terms of features and classifying them using an appropriate classifier, the mental imagery tasks performed can be identified accurately and thus be used for BCI in full applications. The optimal electrodes for mental imagery applications are the C3, Cz and C4 electrodes that are not present in low cost EEG acquisition devices like the Emotiv EPOC+ headset. However, this limitation is overcome in this study. In fact, not all the electrodes available are needed. Most of the information necessary for the mental imagery applications is present at the FC5, FC6, P7, P8, AF3 and AF4 electrodes. Moreover, it is found that the combination of features is able to improve the average cross validation accuracy further. By classifying the Band Power and ApEn features from the electrodes mentioned above using the KNN classifier, an average cross validation accuracy of 99.75% is achieved. If the same features from the FC5, FC6, AF3 and AF4 electrodes are classified, an average cross validation accuracy of 98.55% can be attained. Hence, it is deduced as the best model that meets the aim of this study, requiring only four instead of all the electrodes with a little compromise on the average cross validation accuracy. Based on the model selected, it can be concluded that out of the four mental imagery tasks (LEFT, RIGHT, PUSH and PULL), the PULL mental imagery task is the hardest to be classified, with a classification error of 2.4%. |
| first_indexed | 2025-11-15T19:31:54Z |
| format | Final Year Project / Dissertation / Thesis |
| id | utar-3910 |
| institution | Universiti Tunku Abdul Rahman |
| institution_category | Local University |
| last_indexed | 2025-11-15T19:31:54Z |
| publishDate | 2019 |
| recordtype | eprints |
| repository_type | Digital Repository |
| spelling | utar-39102021-01-08T07:36:50Z Analysis of mental imagery based cognitive tasks for brain computer interface Tang, Chee Hoe T Technology (General) TK Electrical engineering. Electronics Nuclear engineering By representing the EEG signals (brain waves) recorded during mental imagery in terms of features and classifying them using an appropriate classifier, the mental imagery tasks performed can be identified accurately and thus be used for BCI in full applications. The optimal electrodes for mental imagery applications are the C3, Cz and C4 electrodes that are not present in low cost EEG acquisition devices like the Emotiv EPOC+ headset. However, this limitation is overcome in this study. In fact, not all the electrodes available are needed. Most of the information necessary for the mental imagery applications is present at the FC5, FC6, P7, P8, AF3 and AF4 electrodes. Moreover, it is found that the combination of features is able to improve the average cross validation accuracy further. By classifying the Band Power and ApEn features from the electrodes mentioned above using the KNN classifier, an average cross validation accuracy of 99.75% is achieved. If the same features from the FC5, FC6, AF3 and AF4 electrodes are classified, an average cross validation accuracy of 98.55% can be attained. Hence, it is deduced as the best model that meets the aim of this study, requiring only four instead of all the electrodes with a little compromise on the average cross validation accuracy. Based on the model selected, it can be concluded that out of the four mental imagery tasks (LEFT, RIGHT, PUSH and PULL), the PULL mental imagery task is the hardest to be classified, with a classification error of 2.4%. 2019-04-15 Final Year Project / Dissertation / Thesis NonPeerReviewed application/pdf http://eprints.utar.edu.my/3910/1/fyp_EE_2019_TCH.pdf Tang, Chee Hoe (2019) Analysis of mental imagery based cognitive tasks for brain computer interface. Final Year Project, UTAR. http://eprints.utar.edu.my/3910/ |
| spellingShingle | T Technology (General) TK Electrical engineering. Electronics Nuclear engineering Tang, Chee Hoe Analysis of mental imagery based cognitive tasks for brain computer interface |
| title | Analysis of mental imagery based cognitive tasks for brain computer interface |
| title_full | Analysis of mental imagery based cognitive tasks for brain computer interface |
| title_fullStr | Analysis of mental imagery based cognitive tasks for brain computer interface |
| title_full_unstemmed | Analysis of mental imagery based cognitive tasks for brain computer interface |
| title_short | Analysis of mental imagery based cognitive tasks for brain computer interface |
| title_sort | analysis of mental imagery based cognitive tasks for brain computer interface |
| topic | T Technology (General) TK Electrical engineering. Electronics Nuclear engineering |
| url | http://eprints.utar.edu.my/3910/ http://eprints.utar.edu.my/3910/1/fyp_EE_2019_TCH.pdf |