Steady-State Visual Evoked Potentials (SSVEPs) are widely used in spatial selective attention. In this process the two kinds of visual simulators, Light Emitting Diode (LED) and Liquid Crystal Display (LCD), are commonly used to evoke SSVEP. In this paper, the differences of SSVEP caused by these two stimulators in the study of spatial selective attention were investigated. Results indicated that LED could stimulate strong SSVEP component on occipital lobe, and the frequency of evoked SSVEP had high precision and wide range as compared to LCD. Moreover a significant difference between noticed and unnoticed frequencies in spectrum was observed whereas in LCD mode this difference was limited and selectable frequencies were also limited. Our experimental finding suggested that average classification accuracies among all the test subjects in our experiments were 0.938 and 0.853 in LED and LCD mode, respectively. These results indicate that LED simulator is appropriate for evoking the SSVEP for the study of spatial selective attention.
Attention is the cognitive process of selectively concentrating on one area while ignoring others [
The SSVEP can be evoked by different flickers, such as Liquid Crystal Display (LCD), Light Emitting Diode (LED), and Cathode Ray Tube (CRT). The comparison suggested that CRT screen produced more visual fatigue than LCD [
Since both the devices can evoke the SSVEP, the differences in SSEVP results from different simulation environments have been investigated by several researchers like the impact of stimulators in the study of brain computer interface [
In the study of spatial selective attention, the stimulating frequencies must ensure that the responses are unique and strong enough to get the better pattern of results. Aforementioned studies showed that SSVEP would be weak if the stimulus frequency was beyond the range of 5–30 Hz [
In this work, we concentrated on the stimulator selection in SSVEP-based spatial selective attention study. Therefore, the optimal paradigms of LED and LCD were selected, respectively. LED stimulator was presented as LED array in four flickering pattern; one of them was selected as target to enhance subject’s attention. This is a classical paradigm to evoke SSVEP by LED stimulator in the study of spatial selective attention [
A classical left-right dual visual field stimulus paradigm was presented to evoke SSVEP by LCD [
Dual visual field stimulus paradigm by LCD. Subjects monitored the character sequence in one visual field while ignoring the contralateral sequence. Square backgrounds were flickered at 8 Hz in the left field and 12 Hz in the right, which was implemented by two videos of characters in different frequencies. The distance between the monitor and subjects was 60 cm while the view angle was 5 degrees. The subjects were required to press a button as soon as they detect the target “5” to enhance their focus.
Two videos of characters generated in each trial were implemented by E-prime 2.0 (Psychology Software Tools, Inc., version 2.0.8.73). Subjects were directed to sit comfortably in a shielded chamber during the experiment and stared at visual stimulus presented on the LCD 60 cm in front of him/her.
Similar to LCD, in LED experiment, a left-right dual visual field stimulus paradigm was selected based on the classical one [
Dual visual field stimulus paradigm by LED (Miaoqi Electronic, BQ-3401UGC, 505 nm~530 nm; 6000–8000 mcd). The four bars on each side showed the four possible color configurations for each bar. A red circle represented a red LED while the green one represented a green LED. Cue and fixation “+” were displayed by monitor. The distance between the monitor and subjects was 60 cm while the view angle was 5 degrees. Frequency and pattern of LED flicker were controlled by FPGA (CYCLONE IV E, EP4CE15F17C8N) and GRRRG pattern (R means red, G means green) was regarded as the target and subjects were required to push a button once it appears, in order to enhance their focus.
The flicking LEDs form different patterns were RRRRR or GRGRR and so forth (R-red, G-green). In a trial, each pattern lasted for 20 periods. The appearing probabilities of each pattern were 70% for RRRRR and 10% for GRGRR, RRGRG, and GRRRG, respectively.
The target in LCD experiment was character “5,” while the target in in LED experiment was pattern “GRRRG.” At the beginning of each trial, a fixation “+” was shown on the screen; then an arrow will appear near the crossing, indicating which visual field the subject should attend, as shown in Figure
Design of the trial. At the beginning of each trial, an arrow will appear near the crossing, indicating which visual field (left/right) he/she should attend. Sequences and the flickering lasted 12 s in each trial.
Sequences and the flickering lasted 12 s in each trial and there were total 20 trials (10 attend left, 10 right) in the experiment.
In order to investigate the efficiency of LED stimulus, additional experiment was added. This time, the flickering frequencies were changed to 20.8 Hz and 27.8 Hz, which could verify the range (can be set to higher frequency) and accuracy (0.1 Hz in the additional experiment) of the LED stimulator.
The recording room was shielded with a Faraday cage (Changzhou Leining Electromagnetic Shield Equipment Co., Ltd., GPQ2), shielding electromagnetic interference from 14 kHz to 10 GHz. Test equipment was GES 300 System (EGI product; sensor array: 64-channel adult-sized head cap; EEG acquisition software: Net Station; computer: PowerPC G5; amplifier: Net Amps 300). The sample rate was 250 Hz, and the filtering window was 0.3 to 100 Hz.
Data were collected from 8 healthy right-handed volunteers (6 males, 2 females; age: 22 ± 3 years) with normal (or corrected to normal) vision. Most of the processing was performed in Matlab (Version R2013a, MathWorks) and some preprocessing, such as filtering, segmentation, and reference electrode conversion, was executed by Net Station (Version 4.5.7, EGI).
Among the three experiments, the EEG data were filtered through 1–40 Hz band-pass filter. Raw EEG data were segmented to 20 trials (10 attend left, 10 attend right) for each channel. Each epoch was carefully checked and the bad trial was rejected for further analysis, which contained the artifacts related to eye blink, eye movement, and EMG.
Fast Fourier Transform (FFT) technique was adopted for the power spectrum analysis and the topological graph was plotted to show the differences of two types of SSVEP, evoked by LCD and LED, in the frequency domain.
In order to reduce the influence of the nontarget frequency on extracting the feature of SSVEP to a maximum extent, as shown in Figure
(a) The schematic diagram of the train algorithm. Frist, segment the raw data into small pieces (segmented EEG). Second, convolute each segmented EEG with sine signal in both two frequencies, respectively (short SSVEPs). Finally, calculate the short SSVEPs variance as the final feature and training the classifier. (b) The schematic diagram of the test algorithm.
The frequency analysis results in the LED experiment and LCD experiment were shown in Figure
The result of PSD and topological graph. (a) PSD and topological graph of all channels at in LED mode when subject noticed 8 Hz; (b) PSD and topological graph of all channels in LCD mode when subject noticed 8 Hz; (c) PSD and topological graph of all channels in LED mode when subject noticed 12 Hz; (d) PSD and topological graph of all channels in LCD mode when subject noticed 12 Hz.
According to Figure
Position of the recording electrodes. Green indicates the frontal area, and red indicates the occipital area. The electrodes marked with yellow were selected after analysis.
The PSD analysis of subjects was carried out on the results of O1 (CH35),
The result of PSD of O1,
In order to explore the efficiency of LED mode, an additional experiment was added, in which the paradigm was the same as the paradigm shown in Figure
Figures
The result of PSD. (a) PSD of all channels when subject noticed 20.8 Hz; (b) PSD of all channels when subject noticed 27.8 Hz; (c) averaged PSD of O1,
The classification process has been carried out 10 times on each subject’s data and the classification accuracies were shown in Figure
Classification accuracy of two models. Black bars represent the accuracy of LED mode while red bars represent the accuracy of LCD mode. Vertical lines give standard errors.
Classification accuracies of different modes among different time scales. The black solid line represents the accuracy of LED mode while the red dash line represents the accuracy of LCD mode. Vertical lines give standard errors.
This comparison experiment between LCD and LED was carried out and SSVEP signals were both evoked by these two devices with flickers; still there are apparent spectrum frequency differences between attentive and inattentive states of subject; moreover the topological graph in LED mode suggests that the SSVEPs evoked by it have the strong discriminability, which is an important feature when studying the spatial selective attention. Consequently, it is possible to detect the attention distribution of the subject effectively by using LED stimulator.
Furthermore, in LCD mode, a strong frequency component around 10 Hz was observed. It was in alpha band and cannot be easily eliminated. However, in LED mode, this alpha component was not observed. Since the alpha waves are in close relationship with the degree of the attention (as pointed by Dewan [
The result from the additional experiment demonstrated that the LED stimulator could evoke the SSVEP with high frequency precision, which makes it possible to divide the space into more subspaces. Beside this, the flexibility of LED stimulator in both range and accuracy were also tested in this part.
The results obtained from the classification accuracy graph proposed that the SSVEPs evoked by LED mode were easier to distinguish as compared to the LCD mode in all time scales from 0.5 s to 3 s which will contribute to the detection of the subject’s selective attentiveness which enable us to use it in the selective attention as a measurement method with high accuracy.
In this paper, the responses of evoked SSVEP from two different stimulators were studied in detail. The comparison of the two different stimulators was carried out in the analysis of spectrum as well as the accuracy of classification based on two classical paradigms of spatial selective attention proposed by Morgan et al. and Müller et al. with slight modifications. Results showed that strong SSVEP components could be evoked in LED mode as compared to the LCD mode. The best region to acquire the SSVEP was occipital lobe for both modes. The flexibility in generating highly precise and the wide range of stimulating frequency are the advantages of LED mode.
Beside this, there is a significant difference between noticed and unnoticed frequency spectrum while in LCD mode this difference was relatively limited and it was difficult to change the stimulating frequencies. The classification accuracies among all subjects were 0.938 and 0.853 in LED mode and in LCD mode, respectively. This leads us to the conclusion that LED stimulator is an appropriate equipment to study the spatial selective attention and its applications.
All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2008 (5).
Informed consent was obtained from all individual participants included in the study.
The authors declare that they have no conflict of interests.
Songyun Xie and Chang Liu contributed equally to this work.
This work was supported in part by the National Science Fund of China under Grant 61273250, in part by the Fundamental Research Funds for the Central Universities (3102015BJ(II)GH07), in part by graduate starting seed fund of Northwestern Polytechnical University (Z2016124), and in part by Scientific and Technological Plan Project of Shaanxi Province (2015GY003). This work was supported by NPU-TUB Joint Laboratory of Neural Informatics.