دانلود رایگان مقاله انگلیسی طبقه بندی تصورات حرکتی هم کنش گر مغز و کامپیوتر بر اساس الکتروانسفالوگرافی گوش - IEEE 2018

Abstract

Brain-computer interface (BCI) researchers have shown an increased interest in the development of earelectroencephalography (EEG), which is a method for measuring EEG signals in the ear or around the outer ear, to provide a more convenient BCI system to users. However, the ear-EEG studies have researched mostly targeting on a visual/auditory stimulibased BCI system or a drowsiness detection system. To the best of our knowledge, there is no study on a motor-imagery (MI) detection system based on ear-EEG. MI is one of the mostly used paradigms in BCI because it does not need any external stimuli. MI that associated with ear-EEG could facilitate useful BCI applications in real-world. Hence, in this study, we aim to investigate a feasibility of the MI classification using ear-around EEG signals. We proposed a common spatial pattern (CSP)- based frequency-band optimization algorithm and compared it with three existing methods. The best classification results for two datasets are 71.8% and 68.07%, respectively, using the eararound EEG signals (cf. 92.40% and 91.64% using motor-area EEG signals).

IV. DISCUSSION AND CONCLUSION

In this paper, we classified the 2-class MI tasks using the ear-around EEG. And, the performance was compared with that of the motor area to verify the feasibility of the motor imagery classification in the ear-EEG. The proposed method showed better performance than the other methods by finding the optimal frequency-band through the spectral and the temporal filter. However, as expected, ear-EEG based MI classification showed lower accuracies compared with using motor area EEG signals. Note that the performance of the ‘Ear’ with the ‘Motor’, it showed 77.71% and 74.28%, respectively (Table I and II). In this study, we used conventional EEG electrodes attaching around the ear far away approximately 1.5 cm from the ear. However, in future work, we will evaluate the proposed method using ear-EEG electrodes that more close to the ear.