Spectral Graph Theory-Based Spatio-spectral Filters for Motor Imagery Brain–Computer Interface
01 Jan 2021-pp 247-256
TL;DR: A novel approach that utilizes graph theory-based unsupervised feature selection method to determine a reduced set of non-redundant and relevant frequency bands is proposed and shows improvement in classification performance.
Abstract: Motor imagery brain–computer interfaces are one of the widely adopted techniques for imparting basic communication capability to motor disabled patients The preciseness of a motor imagery BCI task classification is highly dependent on identifying the subject-specific relevant subset of frequency filters This article proposes a novel approach that utilizes graph theory-based unsupervised feature selection method to determine a reduced set of non-redundant and relevant frequency bands The empirical analysis of the proposed method is conducted on publicly available datasets, and the obtained results show improvement in classification performance Further, the performed Friedman statistical test also establishes that the proposed approach surpasses the baseline techniques in classification accuracy
References
More filters
TL;DR: A computer‐aided diagnosis (CAD) method using T1‐weighted magnetic resonance imaging (MRI) to differentiate PD from controls is developed.
Abstract: Introduction
Parkinson's disease (PD) is a neurological disorder, which is diagnosed on the basis of clinical history and examination alone as there are no diagnostic tests available. However, the current diagnosis highly depends on the knowledge and experience of clinicians and hence subjective in nature. Thus, the focus of this study is to develop a computer-aided diagnosis (CAD) method using T1-weighted magnetic resonance imaging (MRI) to differentiate PD from controls.
Method:
The proposed method utilizes graph-theory-based spectral feature selection method to select a set of discriminating features from whole brain volume. A decision model is built using support vector machine as a classifier with leave-one-out cross-validation scheme. The performance measures, namely, sensitivity, specificity, and classification accuracy, are utilized to evaluate the performance of the decision model. The efficacy of the proposed method is checked on volumetric 3D T1-weighted (1 mm iso-voxel) MRI dataset of 30 PD patients and 30 age and gender matched controls acquired with 1.5T MRI scanner.
Results:
Experimental results demonstrate that the proposed method is able to differentiate PD from controls with an accuracy of 86.67%, which encourages the use of CAD. The performance of the proposed method outperforms the existing methods except one. In addition, it is observed that the maximum number of selected features belong to caudate region followed by cuneus region. Thus, these regions may be considered as potential biomarkers in diagnosis of PD.
Conclusion
The proposed method may be utilized by the clinicians for diagnosis of PD. © 2015 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 25, 245–255, 2015
27 citations
01 Jan 2016
TL;DR: This paper proposes a method that obtains features from many variable size subbands within a given frequency band using CSP, and Euclidean distance measure is used to obtain the relevant features.
Abstract: Common spatial pattern (CSP) is a commonly used feature extraction technique for motor imagery brain computer interface. CSP provides poor performance when features are extracted from unfiltered or irrelevant frequency band filtered data. In order to overcome this problem, Subband CSP (SBCSP) and Filter Bank CSP (FBCSP) have been proposed in literature to extract features from several fixed size subbands. However, both SBCSP and FBCSP require manually fixing the size of subbands to obtain higher performance. In this paper, we propose a method that obtains features from many variable size subbands within a given frequency band using CSP. Further, Euclidean distance measure is used to obtain the relevant features. The efficacy of the proposed method is evaluated in terms of classification error on BCI Competition III dataset IVa and BCI competition IV dataset Ia. Experimental results demonstrate that the proposed method achieves better performance in comparison to CSP, SBCSP and FBCSP.
15 citations
01 Oct 2019
TL;DR: The work presented in this paper extracts the EEG phase synchrony feature called Phase Lock Value (PLV) to decode Motor Imagery (MI) of center-out hand movement in right and left directions and offers 5.34% improvement in classification accuracy for the 7 best performing subjects compared to the relevant method in literature.
Abstract: Brain-Computer Interface (BCI) systems translate the users intentions coded by brain activity measures into actions through a control signal without using activity of any muscles or peripheral nerves. Usually, in Electroencephalography (EEG) based BCI experiment protocols, different mental tasks are performed to elicit unique brain signal responses, which are recognized by signal processing and machine learning methods. The work presented in this paper extracts the EEG phase synchrony feature called Phase Lock Value (PLV) to decode Motor Imagery (MI) of center-out hand movement in right and left directions. At first, PLV features of all the EEG channel pairs are extracted to detect the level of synchronization corresponding to the directional hand movements. The most significant channel pairs selected from direction-specific EEG signals corresponding to the imagined hand movement showed characteristic changes in PLV features. Mean Percentage Difference of PLV features are calculated and compared in different frequency bands to identify the most discriminative frequency band for the hand movement classification. The extracted PLV features offer 5.34% improvement in classification accuracy for the 7 best performing subjects compared to the relevant method in literature.
14 citations
01 Nov 2018
TL;DR: A novel approach to select a subset of relevant frequency bands using sequential forward feature selection method from a composite filter bank which consists of Prior-known EEG frequency bands and a set of variable size overlapping frequency bands to improve the performance of motor imagery tasks classification.
Abstract: In order to provide basic communication abilities to people with motor disability, motor imagery brain computer interface is one of most widely used technique. In this paper, we present a novel algorithm (Composite Filter bank based stationary CSP) for determining subject as well as task specific discriminative frequency bands for classification of motor imagery tasks. It is noted in the literature that while performing any motor imagery tasks, two major frequency band of EEG spectrum i.e mu (7-12 Hz) as well as beta (12-30 Hz) bands are actively involved. Hence, in most of the literature work EEG signals were filtered using a frequency band of 7-30 Hz usually before using CSP transformation. However, it is possible that some of the frequencies may not provide useful features to distinguish motor imagery tasks. In this paper, we propose a novel approach to select a subset of relevant frequency bands using sequential forward feature selection method from a composite filter bank which consists of Prior-known EEG frequency bands and a set of variable size overlapping frequency bands to improve the performance of motor imagery tasks classification. Experimental results of the proposed work on publicly available datasets validate the effectiveness of the proposed method. Friedman statistical test conducted further shows that the proposed approach significantly outperforms the existing methods.
9 citations
01 Oct 2019
TL;DR: The unsupervised discriminative feature selection (UDFS) is employed here to select the dominant features to improve classification accuracy of motor imagery task acquired by EEG signals.
Abstract: The major challenge in Brain Computer Interface (BCI) is to obtain reliable classification accuracy of motor imagery (MI) task. This paper mainly focuses on unsupervised feature selection for electroencephalography (EEG) classification leading to BCI implementation. The multichannel EEG signal is decomposed into a number of subband signals. The features are extracted from each subband by applying spatial filtering technique. The features are combined into a common feature space to represent the effective event MI classification. It may inevitably include some irrelevant features yielding the increase of dimension and mislead the classification system. The unsupervised discriminative feature selection (UDFS) is employed here to select the subset of extracted features. It effectively selects the dominant features to improve classification accuracy of motor imagery task acquired by EEG signals. The classification of MI tasks is performed by support vector machine. The performance of the proposed method is evaluated using publicly available dataset obtained from BCI Competition III (IVA). The experimental results show that the performance of this method is better than that of the recently developed algorithms.
4 citations