Estimation of driver attention using Visually Evoked Potentials

Visual evoked potential (VEP)-related EEG signals have attracted widespread attention in the construction of brain-computer interface (BCI) systems. However, long-term use of the BCI system is prone to fatigue. Effectively coping with fatigue is the key to expanding BCI application scenarios. In this brief, we conduct two VEP experiments and propose a novel, concise and practical convolutional neural network (CNN) model to study and solve this problem. In detail, firstly, in terms of the stimulus frequencies in VEP experiments, we intercept the spectrum sequences of the fundamental waves and two harmonics. All these spectrum sequences are then paralleled input into the proposed multi-harmonic linkage CNN (MHLCNN) model, through three branches. Finally, the linkage characteristics of multiple harmonics are combined for VEP signal classification. The experimental results demonstrate that the proposed model shows good scalability, and can respectively improve the average classification accuracy by 23.17% and 15.85% when applied to steady-state VEP (SSVEP) signals in fatigue state and steady-state motion VEP (SSMVEP) signals, compared with the benchmark methods. All these provide a feasible and high-quality solution for reducing the impact of fatigue on the performance of VEP-based BCI systems.

In this paper, to develop a hybrid brain-computer interface based on Steady State Visual Evoked Potential(SSVEP) and P300, we validate a SSVEP brain-computer interface (BCI) with the hybrid stimuli of SSVEP and P300, and analyze the effect of P300 visual stimuli on the classification accuracy of the SSVEP BCI. The hybrid stimuli include the SSVEP stimuli, which consist of twochessboardimagespresented on a LCD screen and flashing at particular frequencies, respectively, and the P300 stimuli, which include nine characters and locate between the SSVEP stimuli. We extract the fundamental frequencies and second harmonics of the frequencies of the SSVEP stimulifrom EEG data as features, andapply a Fisher's linear discriminant analysis to build the classifier. The experimental results from three subjects show that the SSVEP potentials can be elicited well under the hybrid stimuli of SSVEP and P300, and the P300 stimuli under the hybrid condition have no significant effect on the SSVEP BCI. This work laysa foundationfor developing a hybrid BCIbased on SSVEP and P300.

543 P120 component of foveal visual evoked potentials encodes luminance contrast in humans

A portable system capable of measuring steady-state visual-evoked potentials (SSVEP) was developed to provide an objective, quantifiable method of electroencephalogram (EEG) testing following a traumatic event. In this study, the portable system was used on 65 healthy rugby players throughout a season to determine whether SSVEP are a reliable electrophysiological biomarker for concussion. Preceding the competition season, all players underwent a baseline SSVEP assessment. During the season, players were re-tested within 72 h of a match for either test-retest reliability or post-injury assessment. In the case of a medically diagnosed concussion, players were reassessed again once deemed recovered by a physician. The SSVEP system consisted of a smartphone housed in a VR-frame delivering a 15 Hz flicker stimulus, while a wireless EEG headset recorded occipital activity. Players were instructed to stare at the screen's fixation point while remaining seated and quiet. Electrodes were arranged according to the 10-20 EEG-positioning nomenclature, with O1-O2 being the recording channels while P1-P2 the references and bias, respectively. All EEG data was processed using a Butterworth bandpass filter, Fourier transformation, and normalization to convert data for frequency analysis. Players SSVEP responses were quantified into a signal-to-noise ratio (SNR), with 15 Hz being the desired signal, and summarized into respective study groups for comparison. Concussed players were seen to have a significantly lower SNR compared to their baseline; however, post-recovery, their SNR was not significantly different from the baseline. Test-retest indicated high device reliability for the portable system. An improved portable SSVEP system was also validated against an established EEG amplifier to ensure the investigative design is capable of obtaining research quality EEG measurements. This is the first study to identify differences in SSVEP responses in amateur athletes following a concussion and indicates the potential for SSVEP as an aid in concussion assessment and management.

Transient visual evoked potentials (VEP) were recorded simultaneously from 16 electrodes evenly placed over posterior scalp locations covering the occipital, posterior parietal and temporal areas. Interhemispheric amplitude difference of the N70 deflection was established across 6 homologous lateral electrode pairs in 15 normal controls and 32 patients with chiasmatic or retrochiasmatic cerebral lesions. Twenty-three of these had known homonymous or bitemporal field defects while 9 had normal fields on routine perimetry. Significant interhemispheric asymmetry of any single electrode pair occurred in 55% of the 32 patients with known pathology, while the cumulative yield of all electrode pairs was over 80 percent. The diagnostic yield of individual electrode pairs was significantly different: the electrode pair placed over the temporo-parietal junction detected the highest number of abnormalities. The cumulative abnormality taken over all lateral electrode pairs could be described with a curve well fitted with a probability summation function. It is inferred that the contribution of several independent generator sources is reflected in the N70 of the pattern onset VEP. The results suggest that multichannel recording of the interhemispheric amplitude distribution of the N70 of the onset VEP is useful for the evaluation of paracentral visual field defects.

Electrophysiological evidence for an early attentional mechanism in visual processing in humans

Central to our understanding of how visual-evoked potentials (VEPs) contribute to visual processing is the question of where their anatomical sources are. Three well-established measures of low-level visual cortical activity are widely used: the first component ("C1") of the transient and multifocal VEP, and the steady-state VEP (SSVEP). Although primary visual cortex (V1) activity has often been implicated in the generation of all three signals, their dominant sources remain uncertain due to the limited resolution and methodological heterogeneity of source modelling. Here, we provide the first characterisation of all three signals in one analytic framework centred on the "cruciform model", which describes how scalp topographies of V1 activity vary with stimulus location due to the retinotopy and unique folding pattern of V1. We measured the transient C1, multifocal C1, and SSVEPs driven by an 18.75 Hz and 7.5 Hz flicker, and regressed them against forward models of areas V1, V2, and V3 generated from the Benson-2014 retinotopy atlas. The topographic variations of all four VEP signals across the visual field were better captured by V1 models, explaining between 2 and 6 times more variance than V2/V3. Models with all three visual areas improved fit further, but complementary analyses of temporal dynamics across all three signals indicated that the bulk of extrastriate contributions occur considerably later than V1. Overall, our data support the use of peak C1 amplitude and SSVEPs to probe V1 activity, although the SSVEP contains stronger extrastriate contributions. Moreover, we provide elaborated heuristics to distinguish visual areas in VEP data based on signal lateralisation as well as polarity inversion.

Spectrum-Enhanced TRCA (SE-TRCA): A novel approach for direction detection in SSVEP-based BCI

BackgroundFlash visual evoked potentials (VEPs) were abnormal in a cohort of 100 neonates exposed to maintenance methadone in utero. This prospective cohort study now describes clinical visual and electrophysiological outcomes...

Objective.With the recent development of visual evoked potential (VEP) based brain-computer interfaces (BCIs), the stimulus paradigm has been continuously innovated, in which the pursuit of higher BCI performance and better user experience has become indispensable.Approach.To optimize the stimulus paradigm, a 12-target online BCI system was designed in this study by adopting flicker for steady-state VEPs, Newton's ring for steady-state motion VEP, and frame rate based video stimulus, respectively. The signal characteristics of VEP, classification accuracy, and user experience of the three stimulus paradigms were quantitatively evaluated and compared.Main results.The online information transfer rates for the three stimulus paradigms were 53.77 bits min-1, 51.41 ± 3.55 bits min-1, and 52.07 ± 3.09 bits min-1, respectively. The video stimulus had a significantly better user experience, while the flicker stimulus showed the worst.Significance.These results demonstrate the advantage of the proposed video stimulus paradigm and have significant theoretical and applied implications for developing VEP-based BCI systems.

Perception of a visual target can strongly deteriorate in the presence of flanking elements (crowding). Crowding is the main limiting factor in reading and a major contributor to poor vision in amblyopia. In crowding, elements appear jumbled and thus are hard to recognize. Classic pooling models propose that crowding occurs via pooling of information from low-level to high-level visual areas. Because of pooling, receptive fields become larger and, at final stages, encompass both the target's and flankers' locations. As a result, the brain is unable to perceive the target separately from the flankers. Pooling models predict that adding flankers or increasing the size of the flankers should always increase crowding strength. In a series of previous studies from our laboratory and other groups, this view was shown to be over-simplistic and incomplete. In many situations, crowding can be reduced and good performance recovered by adding flanking elements or by increasing their size. This can occur when flankers ungroup from the target and group with each other. Grouping between elements is thus a key component in crowding. The neural mechanisms of crowding are unknown at the moment. Here, I used EEG to study the time course of crowding and the brain areas involved in the processing of crowding. Using event-related potentials (ERPs), I showed that crowding is a late visual process. In particular, the N1 component of ERPs (around 190 ms) was suppressed in crowding while earlier components did not correlate with crowding. The N1 suppression was caused by reduced neural activity in high-level visual areas such as the lateral occipital cortex. In a second study, I attempted to replicate the results with gabor stimuli. In a third study, I used steady-state visually evoked potentials (ssVEPs) to disentangle target and flanker processing in crowding. In this study, I showed that the target is selectively suppressed when it groups with the flankers. In addition to these three projects on crowding, I worked on time-frequency analysis that I performed on backward masking data in schizophrenia patients.

This paper presents a novel brain computer interface (BCI) design employing visual evoked potential (VEP) modulations in a paradigm involving no dependency on peripheral muscles or nerves. The system utilizes electrophysiological correlates of visual spatial attention mechanisms, the self-regulation of which is naturally developed through continuous application in everyday life. An interface involving real-time biofeedback is described, demonstrating reduced training time in comparison to existing BCIs based on self-regulation paradigms. Subjects were cued to covertly attend to a sequence of letters superimposed on a flicker stimulus in one visual field while ignoring a similar stimulus of a different flicker frequency in the opposite visual field. Classification of left/right spatial attention is achieved by extracting steady-state visual evoked potentials (SSVEPs) elicited by the stimuli. Six out of eleven physically and neurologically healthy subjects demonstrate reliable control in binary decision-making, achieving at least 75% correct selections in at least one of only five sessions, each of approximately 12-min duration. The highest-performing subject achieved over 90% correct selections in each of four sessions. This independent BCI may provide a new method of real-time interaction for those with little or no peripheral control, with the added advantage of requiring only brief training.

Brain machine interfaces (BMIs) provide one of the important means to interact with external world for a patient who lost motor function by using brain electrical signals such as scalp electroencephalogram (EEG). The main function of BMIs is to detect, analyze and classify the modulation of EEG according to a user's intent and to decode it to demonstrate to others. Most of conventional BMIs with relatively high speed and high classification accuracy have a requisite that a user can open and move his/her eyes toward a target corresponding to user's intent and gaze at it. Locked-in patients, who have lost almost all motor function including facial muscle, cannot control the eye-movement as well as gaze. The present study aimed at developing a BMI which was independent of the eye-movement and eye-gaze as a final goal. In this paper, we discuss the feasibility of binary class of BMI available with eyes-closed condition using a recent our newly observation that the steady-state visual evoked potential (SSVEP) could be modulated by performing the mental tasks even in the eyes-close state. Temporal evolution of the modulation of SSVEP was investigated using the amplitude of SSVEP and Kullback-Leibler (KL) divergence. The amplitudes of fundamental SSVEP (10 Hz) averaged across 11 participants showed significant difference between the mental focusing task and relaxed runs from the onset of the flicker stimuli with the frequency of 10 Hz and the intensity of 5 lx, whereas those of the second harmonic SSVEP (20 Hz) gave only small difference without statistical significance. For the image recalling task, the amplitude differences between the fundamental SSVEPs in the task and relaxed conditions did not show significant difference, whereas the amplitude differences of the second harmonic SSVEP showed the main effect of the image recall task. The KL divergences for the fundamental SSVEP amplitude showed large differences at the onset of flicker stimuli and gradual decrease with time at P4 and O2 for the mental focus task, yet no such characteristic evolution was not seen in KL divergence for the second harmonic SSVEP. On the other hand, the KL divergences during the image recalling were increased with time for both the fundamental and second harmonic SSVEPs at most of the electrode sites. The classification accuracies of the proposed BMI exceeded 75 % in average across the subjects for both tasks using Fisher's linear discriminant analysis.

Spatial frequency masking with the sweep-VEP.

This paper presents an electroencephalogram (EEG) based brain computer interface (BCI) design as a control mechanism for the Open Vibe Space shooter game. BCI using Steady State Visual Evoked Potential (SSVEP) is designed. The SSVEP is generated in response to a flickering visual stimulus. The setup uses different flickering frequencies which are displayed on a Liquid Crystal Display (LCD) monitor to elicit SSVEP responses. The EEG signals are recorded over the visual cortex of the brain. These signals are pre-processed to remove undesired information. To extract features, the power information contained in the corresponding frequency bands are calculated. The feature set is fed as an input to the classifier to identify the target class. This study tried to compare the performance of different classification algorithms in terms of accuracy, information transfer rate and the response time. Based on the outcome, support vector machine classifier with radial basis function recorded the best average accuracy of 90%.