Determination of the Type of The Imagined Movement of Organs in People with Mobility Disabilities Using Corrected Common Spatial Patterns

Subject Areas : Communication

Alireza Pirasteh ¹ (Department of Biomedical Engineering, South Tehran Branch, Islamic Azad University, Tehran, Iran)
Manouchehr Shamseini Ghiyasvand ² (Department of Biomedical Engineering, South Tehran Branch, Islamic Azad University, Tehran, Iran)
Majid Pouladian ³ (Department of Biomedical Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran)

Received: 2022-01-26 Accepted : 2022-02-21 Published : 2022-06-01

Keywords: Brain-Computer interface (BCI), Feature extraction, classifier, EEG Signal Processing, CSP, CSSP,

Abstract :

In order to help people with disabilities, understanding presence of the coronavirus (covid-19) pandemic increasingly highlights the need for emerging technologies. As we know, brain computer interface (BCI) systems were hired to resolve the important challenges on the quality of life of people with disabilities and improve disabled person independent in performing daily activities. Therefore, in this work, BCI systems were furnished to study the type of movement of a person imagines from EEG signals. Before starting to analyze data, frequency bands and brain regions were first associated with motion imaging. Then, various types of spatial and frequency filters were applied to reduce signal noise, after that features were extracted by improving CSP algorithms like CSSP. Because the appropriate frequency band is not selected, the CSP results, which depend on frequency filtering, will not have the desired results, therefore CSSP method based on FIR filters is used. It means that we apply a frequency filter and frequency optimization occurred. The used data is standard data provided on bbci.de. In this database, 9 people have undergone EEG registration. Signal recording was performed in four visual classes including left-hand movement, right-hand movement, both feet, and language. To select the feature, we used the SFS feature algorithm. This algorithm achieved high accuracy by selecting six features together and using SVM classifier. In total, while the accuracy in the CSP method was 87.5%, in the CSSP method it reached 93.6 %.