Connectivity steered graph Fourier transform for motor imagery BCI decoding

Abstract

Objective. Graph signal processing (GSP) concepts are exploited for brain activity decoding and particularly the detection and recognition of a motor imagery (MI) movement. A novel signal analytic technique that combines graph Fourier transform (GFT) with estimates of cross-frequency coupling (CFC) and discriminative learning is introduced as a means to recover the subject’s intention from the multichannel signal. Approach. Adopting a multi-view perspective, based on the popular concept of co-existing and interacting brain rhythms, a multilayer network model is first built from empirical data and its connectivity graph is used to derive the GFT-basis. A personalized decoding scheme supporting a binary decision, either ‘left versus right’ or ‘rest versus MI’, is crafted from a small set of training trials. Electroencephalographic (EEG) activity from 12 volunteers recorded during two randomly alternating, externally cued, MI tasks (clenching either left or right fist) and a rest condition is used to introduce and validate our methodology. In addition, the introduced methodology was further validated based on dataset IVa of BCI III competition. Main results. Our GFT-domain decoding scheme achieves nearly optimal performance and proves superior to alternative techniques that are very popular in the field. Significance. At a conceptual level, our work suggests a fruitful way to introduce network neuroscience in BCI research. At a more practical level, it is characterized by efficiency. Training is realized using a small number of exemplar trials and decoding requires very simple operations that leaves room for real-time implementation.

Covariation Informed Graph Slepians for Motor Imagery Decoding

“…The transition to an online scenario appears feasible, due to the decoder's ability to provide near real-time response. Furthermore, its verified ability to reliably discriminate between resting-state and MI-segments (as shown in section III.C) enables us to envision a self-paced decoding scheme that proceeds in two stages (as presented in our previous study [41]). The first stage will include a "GSL-based switch" with the decoder detecting an MI-event (as deviation from an "idle" brain state), while the second stage will include a "GSL-based discriminator" with the decoder identifying the type of the MI-event (i.e.…”

“…GSP has been successfully adopted by the neuroscientific community, with the earliest approaches providing elegant brain decoding schemes build upon functional neuroimaging signals [37]- [39]. Recently, graph spectral representations have been employed for analyzing EEG data [40] and designing robust BCI decoders [41], [42]. Nevertheless, there are still several aspects of GSP that remain unexplored by the BCI research community and their successful incorporation in the field holds great promise.…”

Single-Trial Kernel-Based Functional Connectivity for Enhanced Feature Extraction in Motor-Related Tasks

“…For this purpose, Figure 3 illustrates the individual classifier performance at each window length, for which the subjects are displayed on the horizontal axis in decreasing order of the CCF accuracy achieved at τ = 2 s (the continuous line outlined in blue color). The rationale for choosing this specific FC estimation case as the baseline is that the cross-correlation coefficient that is presented in Equation (7a) can be directly associated with the conventional Pearson estimate with the simplest interpretation regarding pairwise relationships [50].…”

A Data Augmentation Scheme for Geometric Deep Learning in Personalized Brain–Computer Interfaces

“…Since electroencephalographic activity is a "fuzzy" signal coming from a complex system and governed by the underlying structure of (and the functional connectivity within) the cortical networks, neuroscientists and BCI researchers have started to exploit the recent advances in the domain of graph signal processing [7] so as to incorporate the functional principles of the networked brain within signal analysis and build reliable brain decoding systems [8]- [10]. In this context, geometric deep learning, which collectively refers to adapting and deploying deep learning on data manifolds, graph patterns and signals registered over irregular grids, could significantly enhance the performance in existing BCI protocols and implementation pipelines.…”