Performance of a Steady-State Visual Evoked Potential and Eye Gaze Hybrid Brain-Computer Interface on Participants With and Without a Brain Injury

Abstract

The brain-computer interface (BCI) and the tracking of eye gaze provide modalities for human-machine communication and control. In this paper we provide the evaluation of a collaborative BCI and eye gaze approach, known as a hybrid BCI. The combined inputs interact with a virtual environment to provide actuation according to a four-way menu system. Two approaches were evaluated: (1) steady state visual evoked potential (SSVEP) BCI with on-screen stimulation; (2) hybrid BCI, which combined eye gaze and SSVEP for navigation and selection. A study comprised participants without known brain injury (non-BI, N=30) and participants with known brain injury (BI, N=14). Twenty-nine of the thirty non-BI participants could successfully control the hybrid BCI, while nine out of the fourteen BI participants were able to achieve control, as evidenced by task completion. The hybrid BCI provided a mean accuracy of 99.84% in the cohort of non-BI participants and 99.14% in the cohort of BI participants. Information transfer rates were 24.41 bits/min in non-BI participants and 15.87 bits/min in BI participants. The research goal is to quantify usage of SSVEP and ET approaches in cohorts of non-BI and BI participants. The hybrid was the preferred interaction modality for most participants for both cohorts. When compared to non-BI participants, it was encouraging that nine out of fourteen participants with known brain injury could use the hBCI technology with equivalent accuracy and efficiency, albeit with slower transfer rates. Index Terms-Brain-computer interface, brain injury, eye tracking, data fusion, virtual environment I. INTRODUCTION isorders that affect the cerebrum and the central nervous system, e.g. stroke, and traumatic brain injury incapacitate the individual concerned, often affecting their mobility and communication. The Future BNCI roadmap estimated stroke prevalence at 3% of the population, and that 250,000 Americans were living with spinal cord injury (47% of whom were quadriplegic) [1]. Rehabilitation and long-term care may

EEG hybrid brain-computer interfaces: A scoping review applying an existing hybrid-BCI taxonomy and considerations for pediatric applications

“…One study had a participant aged 18 years (Kaongoen and Jo, 2017) but most studies had at least one participant within the range of 20-30 years, except for Nann et al (2020) who worked with tetraplegic participants aged 51.8 ± 15.2 years. Only two papers included participants above 40 years of age (Brennan et al, 2020; Nann 2016) were omitted from the graph because their average accuracies were less than 70% (44.5 and 60%, respectively).…”

“…The study reported that he was able to control a robotic hand via motor imagery (MI), even though his accuracy was slightly lower than the average of the other neurotypical participants (76.03% compared to 80.65%). Brennan et al (2020) included 14 participants with brain injury (although only nine completed the hBCI trials) with an average age range of 41.6±13.9 years. Participants underwent trials with both an SSVEP BCI and an SSVEP Eye Tracker hBCI for comparison.…”

“…The multi-physiological signals were mostly EEG and EOG, but Lin et al (2016) combined EEG andEMG andZhang et al (2019a) combined EEG, EOG, and EMG. As for the ones with external input, Mannan et al (2020) and Brennan et al (2020)…”

“…Inputs such as eye trackers, joysticks and switches add reliability to the system and therefore increase its performance. Three of the selected papers used external inputs, Mannan et al (2020) and Brennan et al (2020) used NIR eye trackers and Saravanakumar and Reddy (2019) utilized a camera-based eye tracker. All the systems used the eye gaze as a means to narrow down the possible targets.…”

Asynchronous c-VEP communication tools—efficiency comparison of low-target, multi-target and dictionary-assisted BCI spellers

“…A low number of open Rhine-Waal University of Applied Sciences, Technology and Bionics, 47533 Kleve, Germany. * email: ivan.volosyak@ hochschule-rhein-waal.de stimulus classes are sufficient; spellers with only four or five different flickering patterns are quite common 4,[12][13][14][15] . While low-target spellers allow high classification accuracies, the overall spelling speed is limited, as several selection intervals (typically consisting of stimulation intervals and flicker-free intervals for gaze-shifting, where users can shift their gaze to the next target) are required for letter selections.…”