Single trial classification of many imagery tasks for BCI  

Oleh:

Matsuda, Jumpei ; Sanno, Satoshi ; Misawa, Tadanobu ; Hirobayashi, Shigeki

Jenis: Article from Proceeding
Dalam koleksi: The 14th Asia Pacific Industrial Engineering and Management Systems Conference (APIEMS), 3-6 December 2013 Cebu, Philippines, page 1-7.
Topik: Brain-computer interface (BCI) ; Motor imagery (MI) ; Electroencephalogram (EEG) ; Single-trial classification
Fulltext: 5006_Oyabu.pdf (221.34KB)

Isi artikel

It has recently become possible for us, to use high-performance and low-priced electroencephalography (EEG) due to progress that has been made in neuroscience and signal processing. A non-invasive brain-computer interface (BCI) allows for direct communication between humans and computers by analyzing electrical brain activity, which is recorded on the surface of the scalp using EEG. EEG-based BCIs often utilize recognition of a state of less with motor imagery. This study proposes a Kanji writing imagery task as a new BCI task, and demonstrates its performance, along with conventional tasks (finger or foot tapping, hand grasping, an auditory imagery task, etc.), using 14 electrodes on five healthy subjects. Moreover, we classify the differences in the state of the imagery for each of the same tasks into 4 classes, and demonstrate the utility of these classifications. More specifically, these tasks involved imaging of the hand and the complexity of Kanji in the Kanji writing task, imagining the hand and frequency of tapping in a finger-tapping task, and imaging the hand and the strength of the grasp in a hand-grasping task. When performing the proposed task, the subjects wrote imagery using complex or easy Kanji using their left or right hand (the left hand with easy Kanji, the left hand with complex Kanji, the right hand with easy Kanji, and the right hand with complex Kanji). We classified all combinations of four categories of the two tasks using t-test and the support vector machine (SVM) on all imagery tasks. The results of our experiment, we verify that the efficacy of the proposed task was similar to that of conventional tasks, as well as to indicate a new approach to a BCI task. In the future, we will test the classification of different combinations of these tasks.

Opini Anda

Klik untuk menuliskan opini Anda tentang koleksi ini!

Kembali