Knowledge Vault 3/52 - G.TEC BCI & Neurotechnology Spring School 2024 - Day 5
Wireless EEG recordings and sports
Francisco Fernandes, g.tec medical engineering, Schiedlberg (AT)
<Resume Image >

Concept Graph & Resume using Claude 3 Opus | Chat GPT4 | Llama 3:

graph LR classDef sports fill:#f9d4d4, font-weight:bold, font-size:14px; classDef eeg fill:#d4f9d4, font-weight:bold, font-size:14px; classDef challenges fill:#d4d4f9, font-weight:bold, font-size:14px; classDef wireless fill:#f9f9d4, font-weight:bold, font-size:14px; classDef studies fill:#f9d4f9, font-weight:bold, font-size:14px; classDef artifacts fill:#d4f9f9, font-weight:bold, font-size:14px; A[Francisco Fernandes] --> B[EEG recordings and
sports presentation. 1] A --> C[Reasons: physiological responses,
comparisons, performance. 2] A --> D[EEG markers: ERPs,
imagery, spectral. 3] A --> E[Challenges: artifacts, biomarkers,
limitations. 4] E --> F[Low-movement sports
were chosen. 4] A --> G[Unicorn allows free
movement recording. 5] G --> H[Artifact removal
tools beneficial. 5] A --> I[First study: static
bike, GUSB. 6] A --> J[2022: martial arts,
wireless setup. 7] J --> K[Device, backpack,
tablet, laptop. 8] A --> L[2023: badminton, alpha,
concentration. 9] L --> M[D2 test, real-time
feedback. 10] A --> N[2024: unicorn, hybrid
electrodes tested. 11] N --> O[Wireless streaming,
robust PCs. 11] N --> P[Audio EPs, clear
P300 peaks. 12] A --> Q[Setup: free performance,
signal verification. 13] Q --> R[Audio from
loudspeakers. 13] A --> S[Simulink: collection, transmission,
processing. 14] S --> T[Raw vs artifact-
corrected data. 15] A --> U[Exercises: squats, pushups,
headstands. 16] U --> V[Cleaner corrected data
vs raw. 16] A --> W[Backflip: raw obliterated,
corrected clean. 17] A --> X[Eyes closed: alpha
in raw, corrected. 18] X --> Y[OSCAR in hardware,
real-time. 18] A --> Z[Audio EPs during
movements planned. 19] Z --> AA[Two PCs: worn,
stationary. 19] Z --> AB[In-ear or
loudspeaker options. 20] A --> AC[Visual/tactile EPs possible
but complex. 21] A --> AD[Technical glitches, prior
recording shown. 22] A --> AE[Raw EPs compromised,
OSCAR enables motion. 23] A --> AF[Future wireless EEG,
real-time removal. 24] A --> AG[Questions: treadmill EEG,
YouTube, certificate. 25] A --> AH[eSports: pro gamer
vs athlete EEG. 26] A --> AI[Removal: adaptive filtering,
oversampling. 27] AI --> AJ[Simultaneous ERPs may
be problematic. 27] A --> AK[Motor imagery: athletes
show higher accuracy. 28] A --> AL[Brain-controlled robot
painting by Ilic. 29] A --> AM[Barter: BCI for painting,
borrowing agreement. 30] class A,B,I,J sports; class C,D,E,F,L,M,X,AC,AK eeg; class G,H,N,O,P,Q,R,Z,AA,AB,AF,AI,AJ wireless; class S,T,U,V,W,AD,AE artifacts; class Y,AH studies;


1.- Francisco Fernandes is presenting on EEG recordings and sports, a growing area of interest due to advancements in wireless EEG technology.

2.- Reasons to combine EEG and sports include measuring physiological responses during physical activity, comparing athletes to amateurs, and predicting/improving performance.

3.- Typical EEG markers used in sports studies are ERPs, motor imagery, spectral changes, sleep studies, respiration, GSR, EMG, EOG, and eye tracking.

4.- Challenges in sports EEG studies include movement artifacts, lack of concrete biomarkers/protocols, and technology limitations, so low-movement sports were often chosen.

5.- Wireless devices like unicorn allow free movement during EEG recording. Artifact removal tools are also beneficial for dealing with movement artifacts.

6.- One of the first sports EEG studies 7 years ago used a static bike, GUSB amplifier, and audio evoked potentials via headphones.

7.- In 2022, a wireless setup was used with a subject doing martial arts while wearing a GENO TILO's wireless device.

8.- Setup included EEG device, backpack, tablet streaming data via Wi-Fi, and a laptop providing audio stimulation and receiving data via UDP.

9.- In 2023, a badminton EEG study looked at alpha activity and concentration using G-nautilus 8 channels, D2 test, and real-time feedback.

10.- D2 test results were used to detect concentration levels from alpha/theta during badminton. Bar feedback indicated concentration level in real-time.

11.- For 2024, a unicorn with hybrid electrodes was tested, streaming data wirelessly from a tablet PC to a 2nd robust PC.

12.- Audio evoked potentials were elicited and data was streamed via LSL. Clear P300 peaks were seen despite movement, even with dry electrodes.

13.- Setup allows athletes to freely perform weight lifting, calisthenics etc. while verifying EEG signal quality. Audio stimuli can come from loudspeakers.

14.- Two Simulink models are used - one for unicorn data collection/LSL transmission, one for receiving data/triggers and processing (filtering, artifact correction).

15.- Raw data is shown on the left, artifact-corrected data on the right using OSCAR toolbox. Very clean data is obtained after correction.

16.- Various exercises are performed (squats, pushups, headstands) showing much cleaner data on the right after artifact removal compared to raw data.

17.- A backflip is performed, obliterating the raw EEG but the artifact-corrected signal still looks clean, demonstrating the effectiveness of the approach.

18.- Eyes closed shows clear alpha boost in both raw and corrected data. OSCAR artifact removal is embedded in the hardware for real-time use.

19.- Audio evoked potentials will be elicited while Martin performs movements. Two PCs can be used - one worn by subject, one stationary.

20.- For audio EPs, in-ear headphones are best but loud speakers are an option if headphones are too restrictive for the athletic activity.

21.- Visual/tactile stimuli for EPs are possible in sports studies but more complicated since vision is often required. Auditory is more feasible.

22.- Some technical glitches encountered trying to record EPs in real-time during the live demo. Prior day's recording is shown instead.

23.- Without artifact removal, raw EP averaging is severely compromised by movement. OSCAR provides clear EPs in real-time during motion.

24.- Future opportunities abound for wireless EEG and real-time artifact removal in sports science studies and applications. More spring school demos planned.

25.- Questions are addressed: Reproducing leg movement from treadmill running EEG is feasible. Visit YouTube for missed live session recordings to earn certificate.

26.- Comparing EEG of pro gamers to traditional pro athletes would be an interesting eSports study to examine differences in elite brains.

27.- The artifact removal combines adaptive filtering with steep anti-aliasing and oversampling. Simultaneous visual/auditory ERPs in a session may be problematic.

28.- Motor imagery shows higher accuracy in athletes vs non-athletes, likely due to their greater practice with imagined action for sports performance.

29.- The painting behind the presenters was made by Dragan Ilic using a brain-controlled robot during a 5-day Ars Electronica event.

30.- Christoph bartered a BCI system to Dragan in exchange for the large painting, which Dragan can borrow back for exhibitions as needed.

Knowledge Vault built byDavid Vivancos 2024