A brain-computer interface to control a prosthetic hand / Yong Xin Yi
A Brain-Computer Interface (BCI) system was developed to operate a prosthetic hand and other devices. The Electroencephalogram (EEG) signals were recorded over the sensorimotor cortex area during foot, left or right hand motor imagery. Only two mental tasks and one or two EEG bipolar channels wer...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2005
|
| Subjects: | |
| Online Access: | http://studentsrepo.um.edu.my/7487/ http://studentsrepo.um.edu.my/7487/1/Binder1.pdf |
| _version_ | 1848773403829338112 |
|---|---|
| author | Yong, Xin Yi |
| author_facet | Yong, Xin Yi |
| author_sort | Yong, Xin Yi |
| building | UM Research Repository |
| collection | Online Access |
| description | A Brain-Computer Interface (BCI) system was developed to operate a prosthetic
hand and other devices. The Electroencephalogram (EEG) signals were recorded over
the sensorimotor cortex area during foot, left or right hand motor imagery. Only two
mental tasks and one or two EEG bipolar channels were identified and used in the
online experiments. Autoregressive (AR) modeling was used to extract the features
from the spontaneous EEG signals and Linear Discriminant Analysis (LDA) was used
as the classifier.
Six subjects participated in the online study. However, only three subjects had
sufficient control to proceed to the final application phase. The online classification
errors for these subjects ranged between zero and 17.8% in the subject-training phase.
In the application phase, the subjects were required to complete a preprogrammed test
sequence. The optimal time to complete the test sequence is appr9ximately 6 minutes.
The times taken by the subjects to complete the test sequence were between 8 minutes
20 seconds and 17 minutes. The unintended activations per minute generated by the
subjects varied from zero to 0.8 per minute.
In the present application, high classification accuracy with low unintended
activations is more important than a high information transfer rate (ITR). By
introducing thresholds in the LDA classification rule and averaging the LDA outputs
over 5 seconds to arrive at a decision, we minimize the unintended activations although
the true positives (TP) and the ITR were reduced. The results of the present study show
that three of the subjects were able to use the BCI system to control a prosthetic hand
and other devices. |
| first_indexed | 2025-11-14T13:41:52Z |
| format | Thesis |
| id | um-7487 |
| institution | University Malaya |
| institution_category | Local University |
| last_indexed | 2025-11-14T13:41:52Z |
| publishDate | 2005 |
| recordtype | eprints |
| repository_type | Digital Repository |
| spelling | um-74872017-08-04T05:35:12Z A brain-computer interface to control a prosthetic hand / Yong Xin Yi Yong, Xin Yi T Technology (General) A Brain-Computer Interface (BCI) system was developed to operate a prosthetic hand and other devices. The Electroencephalogram (EEG) signals were recorded over the sensorimotor cortex area during foot, left or right hand motor imagery. Only two mental tasks and one or two EEG bipolar channels were identified and used in the online experiments. Autoregressive (AR) modeling was used to extract the features from the spontaneous EEG signals and Linear Discriminant Analysis (LDA) was used as the classifier. Six subjects participated in the online study. However, only three subjects had sufficient control to proceed to the final application phase. The online classification errors for these subjects ranged between zero and 17.8% in the subject-training phase. In the application phase, the subjects were required to complete a preprogrammed test sequence. The optimal time to complete the test sequence is appr9ximately 6 minutes. The times taken by the subjects to complete the test sequence were between 8 minutes 20 seconds and 17 minutes. The unintended activations per minute generated by the subjects varied from zero to 0.8 per minute. In the present application, high classification accuracy with low unintended activations is more important than a high information transfer rate (ITR). By introducing thresholds in the LDA classification rule and averaging the LDA outputs over 5 seconds to arrive at a decision, we minimize the unintended activations although the true positives (TP) and the ITR were reduced. The results of the present study show that three of the subjects were able to use the BCI system to control a prosthetic hand and other devices. 2005-11-27 Thesis NonPeerReviewed application/pdf http://studentsrepo.um.edu.my/7487/1/Binder1.pdf Yong, Xin Yi (2005) A brain-computer interface to control a prosthetic hand / Yong Xin Yi. Masters thesis, University of Malaya. http://studentsrepo.um.edu.my/7487/ |
| spellingShingle | T Technology (General) Yong, Xin Yi A brain-computer interface to control a prosthetic hand / Yong Xin Yi |
| title | A brain-computer interface to control a prosthetic hand / Yong Xin Yi |
| title_full | A brain-computer interface to control a prosthetic hand / Yong Xin Yi |
| title_fullStr | A brain-computer interface to control a prosthetic hand / Yong Xin Yi |
| title_full_unstemmed | A brain-computer interface to control a prosthetic hand / Yong Xin Yi |
| title_short | A brain-computer interface to control a prosthetic hand / Yong Xin Yi |
| title_sort | brain-computer interface to control a prosthetic hand / yong xin yi |
| topic | T Technology (General) |
| url | http://studentsrepo.um.edu.my/7487/ http://studentsrepo.um.edu.my/7487/1/Binder1.pdf |