Knowledge Vault 3/80 - G.TEC BCI & Neurotechnology Spring School 2024 - Day 9
mindBEAGLE assessment, communication, predication, rehabiliation
Ren Xu, g.tec medical engineering GmbH (AT)
<Resume Image >

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

graph LR classDef assessment fill:#f9d4d4, font-weight:bold, font-size:14px; classDef hardware fill:#d4f9d4, font-weight:bold, font-size:14px; classDef paradigm fill:#d4d4f9, font-weight:bold, font-size:14px; classDef demo fill:#f9f9d4, font-weight:bold, font-size:14px; A[Ren Xu] --> B[demonstrates
MindBeagle system 1] A --> C[MindBeagle: EEG, auditory/tactile
stimuli, real-time 2] C --> D[Assessments enable communication,
predict outcomes 3] C --> E[Hardware: EEG headset,
headphones, stimulators 4] E --> F[Cube integrates hardware,
delivers stimuli 5] C --> G[Implements 6 paradigms:
AEP, VT2/3/7, MI, resting 6] G --> H[Assess consciousness:
enable yes/no communication 7] D --> I[Assessment pages:
EEG, stimuli, ERPs 8] D --> J[Quick test: moving bar,
overall success 9] D --> K[Communication: type questions,
focus no/yes 10] D --> L[Short durations:
DOC patients tire easily 11] D --> M[Results: ERPs, accuracy,
historical data, MI 12] B --> N[Live demo: EEG cap,
signal quality 13] N --> O[AEP: count high tones,
59% accuracy 14] N --> P[Quick AEP unsuccessful,
prior 95% accuracy 15] N --> Q[VT2: count left wrist,
100% accuracy 16] Q --> R[VT2 quick test successful,
matches assessment 17] N --> S[VT3: ankle distractors,
100% accuracy 18] N --> T[Communication: likes MindBeagle yes,
Ren speaking Chinese no 19] N --> U[VT7 assessment and
communication shown briefly 20] N --> V[Motor imagery: left/right hand,
tactile precursors 21] C --> W[Dry EEG possible
with maintained quality 22] G --> X[VT2 easiest for DOC,
short duration 23] G --> Y[Auditory, MI challenging,
longer duration 24] A --> Z[Portable system,
enables home use 25] class B,N,O,P,Q,R,S,T,U,V demo; class C,D,I,J,K,L,M,W assessment; class E,F hardware; class G,H,X,Y paradigm;


1.- Ren Xu gives a practical demonstration of using the MindBeagle system for assessment, communication, prediction and evaluation of DOC patients.

2.- MindBeagle uses EEG to measure brain activity while delivering auditory and tactile stimuli. The computer analyzes EEG data in real-time.

3.- Successful assessments enable yes/no communication with patients. Results reflect brain functions and can predict outcomes. Repeated stimulation may improve function.

4.- MindBeagle hardware includes a 16-channel wireless EEG headset, headphones for auditory stimuli, and 7 vibro-tactile stimulators on the body.

5.- The cube integrates the hardware components and connects to the laptop. It delivers auditory and tactile stimuli and records EEG.

6.- MindBeagle implements 6 paradigms: auditory evoked potentials, vibro-tactile with 2/3/7 stimulators, motor imagery, and resting state EEG.

7.- All paradigms assess consciousness. AEP and VT2 allow quick real-time assessments. VT3, VT7 and MI enable yes/no communication when successful.

8.- Assessment pages show real-time EEG, electrode map and signal quality, stimuli setup, and ERPs. Color indicates stimulation side.

9.- Quick test pages show a moving bar for correct classification of each trial, determining overall success at the end.

10.- Communication pages allow typing questions. Patients focus left for "no", right for "yes". Results appear after completing the question block.

11.- Paradigm durations are kept short (2.5-4 min) as DOC patients tire easily. Quick tests and communication are even shorter (0.5-1.5 min).

12.- Result pages show ERPs, significance, scalp maps, accuracy over averaging, and historical data. MI results show power changes during imagined movement.

13.- Live demonstration begins with Ren preparing the EEG cap on his colleague Vaishali. Signal quality is checked in the software.

14.- In the AEP assessment, Vaishali counts high tones and ignores low tones for 4 minutes. Accuracy is 59% due to some distraction.

15.- Quick AEP test is unsuccessful, likely due to imperfect signal-to-noise ratio and accuracy in the assessment. Prior data showed 95% accuracy.

16.- VT2 assessment has Vaishali count left wrist tactile stimuli for 2.5 min. 100% accuracy is achieved with high quality ERPs.

17.- VT2 quick test is successful with real-time short trial data matching the assessment well. High assessment accuracy enables successful quick tests.

18.- VT3 assessment adds ankle distractors. Vaishali achieves 100% accuracy counting right then left wrist stimuli with very high quality ERPs.

19.- Two communication runs ask Vaishali if she likes MindBeagle (yes - right) and if Ren is speaking Chinese (no - left). Both succeed.

20.- VT7 assessment and communication paradigms are briefly shown, similar to VT3 but with more distractors on the vibro-tactile stimulators.

21.- Motor imagery paradigm is shown, with assessment and communication based on imagining left/right hand movements. Tactile precursors improve accuracy.

22.- In response to a question, Ren states that dry EEG electrodes could work if signal quality is maintained, perhaps using a hybrid approach.

23.- Ren suggests vibro-tactile paradigms, especially VT2, are easiest for DOC patients due to the short duration enabling assessment within limited windows of responsiveness.

24.- Auditory and motor imagery paradigms are more challenging for DOC patients because of the longer duration, during which patients may fall asleep.

25.- The entire MindBeagle system fits in a backpack, enabling use with patients at home. The team traveled from Schiedeberg for this demonstration.

Knowledge Vault built byDavid Vivancos 2024