1. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8, 2010

2. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Outline Brain-Computer Interfaces (BCIs) Steady-state visual evoked potential (SSVEP) SSVEP-based BCIs Stimulation properties Conclusion Questions 2

3. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Brain-Computer Interface A brain-computer interface (BCI) is a system that allows a user to communicate his intent to a system without using any peripheral output pathways. 3

4. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 PerformanceComfort ITR = C * log N / TQuestionnaire Measures of success 4

5. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 BCI Types BCIs can be based on different neuromechanisms. –Sensorimotor activity –P300 / oddball paradigm –Visual evoked potentials BCIs based on the steady-state visual evoked potential appear to be especially promising. 5

6. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Steady-state visual evoked potential Oscillatory light intensity (15 Hz) tVEP (300ms) SSVEP 6

7. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Stimulation 7

8. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Goal To determine the optimal stimulation properties to improve the performance and comfort of SSVEP-based BCIs. Frequency Framerate Stimulation device Environment Contrast Color Spatial frequency Size 8

9. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 SSVEP-based BCI: maze navigation BioSemi 8 channels + 1 reference BCI 2000 Neurostim 6 subjects

10. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 The strength of the SSVEP response depends on the stimulation frequency. Higher frequencies are more comfortable to look at and are less likely to induce epileptic seizures. Frequency Stimulation Frequency (Hz) SSVEP Amplitude 5 10 1215172022252730 35404760 Low frequency region Middle frequency region High frequency region 10

11. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Framerate Which frequencies can be rendered depends on the framerate of the stimulation device (and the number of intermediate states required for the waveform). Assuming that the minimum of 2 states are required (on/off), the device can render all frequencies 1/kF, where F is the framerate and k is any integer > 2. If both states should be displayed an equal amount of time (duty cycle 50%), k needs to be an even integer > 2. 11

12. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Framerate energy of stimulation frequency time (s) S4 18Hz frequency (Hz) power 12

13. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Framerate 13

14. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Display device The most often used display devices are light emitting diodes (LEDs), cathode ray tubes (CRTs) and liquid crystal displays (LCDs). 14

15. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Stimulation devices 15

16. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Environment Noise and light can have an effect on concentration and on how the stimulation is perceived. Frequencies of light and sound in the environment could interfere with the SSVEP response. In the dark, the eye becomes more sensitive to luminance changes. 16

17. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Luminance and contrast The contrast of the stimulation is the most important factor in how noticeable the changes are. More noticeable flicker elicits stronger responses, but be less comfortable. 17

18. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Color Different colors are processed slightly differently in the eyes and brain. Instead of varying luminance, hue and saturation can also be alternated. Green-blue stimulation works really well. Colored backgrounds don’t. 18

19. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Pattern reversal The checkerboard is a rectangle that is made up of smaller rectangles with alternating colors that switch around at every alternation. Some people say that using a checkerboard pattern elicits a larger SSVEP response. Contrary to the case where there is no pattern, the luminance does not have to change when the number of checks of both colors is equal. Also, the SSVEP response is at the alternation rate, or twice the cycle frequency, of the stimulus. 19

20. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Pattern reversal and spatial frequency The spatial frequency of a stimulus defines how many times the pattern is repeated in a certain space. A higher spatial frequency results in more and smaller cells, and has been linked to low contrast stimuli. Higher spatial frequencies were more comfortable, but the relation between spatial frequency and performance is nonlinear. Performance of pattern reversal was worse than single graphic stimulation. 20

21. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Size The larger the stimulus, the more easy it is to see. This can increase SSVEP strength, both when the stimulus is attended to and when it’s not. This means there is an optimal stimulus size, that is likely also influenced by target spacing. 21

22. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Conclusions Stimulation properties can have a significant impact on BCI performance and user comfort. Generally stimuli with a higher contrast are less comfortable, but result in better BCI performance. Stimuli that had good results on both measures are green/blue, gray and blue stimulation on a black background (with red and green coming close). 22

23. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Products Made BCI (demo) Publications: –D. Zhu, J. Bieger, G. Garcia Molina, and R. M. Aarts (2009), A survey of stimulation methods used in SSVEP-based BCIs. Journal of Computational Intelligence and Neuroscience –T. Tsoneva, J. Bieger, and G. Garcia Molina (submitted for publication), Towards error-free interaction, Proceedings of the 32 nd EMBC –J. Bieger, D. Zhu, and G. Garcia Molina (submitted for publication), Effects of Stimulation Properties in Steady-State Visual Evoked Potential Based Brain-Computer Interfaces, Proceedings of the 32 nd EMBC Invention Disclosure proposals: –Affective Categorization of Media Content Using Visual Evoked Potentials –Increased Throughput of Automatic Speech Recognition Systems using Error Detection from Brain signal 23

24. Stimulation Effects in SSVEP-based BCIs Jordi Bieger, July 8 2010 / 24 Questions 24