Received November 30, 2020, accepted December 19, 2020, date of publication January 5, 2021, date of current version January 13, 2021. Digital Object Identifier 10.1109/ACCESS.2021.3049191 Trial Regeneration With Subband Signals for Motor Imagery Classification in BCI Paradigm MD. KHADEMUL ISLAM MOLLA 1 , (Member, IEEE), SANJOY KUMAR SAHA 2 , SABINA YASMIN 3 , MD. RABIUL ISLAM 4 , (Member, IEEE), AND JUNGPIL SHIN 5 , (Senior Member, IEEE) 1 Department of Computer Science and Engineering, University of Rajshahi, Rajshahi 6205, Bangladesh 2 Department of Computer Science and Engineering, Begum Rokeya University, Rangpur 5404, Bangladesh 3 Department of Computer Science and Engineering, Varendra University, Rajshahi 6204, Bangladesh 4 Institute of Global Innovation Research, Tokyo University of Agriculture and Technology, Tokyo 184-8588, Japan 5 School of Computer Science and Engineering, The University of Aizu, Fukushima 965-8580, Japan Corresponding authors: Md. Khademul Islam Molla (khademul.cse@ru.ac.bd) and Jungpil Shin (jpshin@u-aizu.ac.jp) ABSTRACT Electroencephalography (EEG) captures the electrical activities of human brain. It is an easy and cost effective tool to characterize motor imager (MI) task used in brain computer interface (BCI) implementation. The MI task is represented by short time trial of multichannel EEG. In this paper, the raw EEG trial is regenerated using narrowband signals obtained from individual channel. Each channel of EEG trial is decomposed into a set of subband signals using multivariate discrete wavelet transform. The selected subbands are organized in two different ways namely vertical arrangement of subbands (VaS) and horizontal arrangement of subbands (HaS) to regenerate the trials. The features are extracted from each of the arrangements using common spatial pattern (CSP). An optimum number of features are used to classify the motor imagery tasks represented by EEG trials. The effectiveness of two classifiers – linear discriminant analysis (LDA) and support vector machine (SVM) are studied. The performances of the proposed methods are evaluated using publicly available benchmark datasets. The experimental results show that it performs better than the recently developed algorithms. INDEX TERMS Brain computer interface (BCI), discrete wavelet transformation, electroencephalography (EEG), motor imagery, narrowband signals, subband decomposition. I. INTRODUCTION Brain-computer interface (BCI) is a relatively new tech- nology that helps in motor rehabilitation and neuromuscu- lar disorder of the paralyzed patients. It is a tool which offers a communication channel and control capabilities that allow direct connection between human brain and external device using brain activities [1]. The technology used in BCI translates human intention into command which cannot pass through the brain’s normal output pathway of periph- eral nerves and muscles [2]. Recently, several attempts are reported to achieve the goal [3] of BCI. The electroen- cephalography (EEG) based BCI system is cost effective and easy to implement with minimal clinical risk as well as invasive in nature [4], [5]. It captures the brain activities as an electrical signal from the scalp using a set of electrodes. Motor imagery (MI) is a common mental task that is widely used in BCI implementation [6]. In MI based BCI, a subject is The associate editor coordinating the review of this manuscript and approving it for publication was Zhanspeng Jin . required to perform an imagination in the brain corresponding to a specific mental task. The recorded EEG signals related to MI are classified to translate into corresponding control com- mand for different imaginary tasks like movement of hand, foot etc. [7]. In terms of neurophysiology, motor imagery accompanies attenuation or enhancement of rhythmical syn- chrony over the sensorimotor cortex with the frequency bands of alpha (8 – 13 Hz) and beta (14 – 30 Hz) [8]–[11]. This paper focuses on EEG based classification of two motor imagery tasks. The most of the current MI based BCIs have three stages including pre-processing, feature extraction and classification [12]. To improve its classification accuracy, the pre-processing is an important step in which the unwanted components of signals are suppressed. The real world mul- tichannel EEG signal is non-stationary. The subband based approach to EEG classification performs better than that of full band method. Fourier transform (FT) based filtering is usually used for subband decomposition of EEG signal. Multiband tangent space mapping is used in [13] to extract the 7632 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ VOLUME 9, 2021