The End Of Knowledge - Vault 3 - G.TEC BCI & Neurotechnology Spring School 2024 - Day 8

graph LR classDef recovery fill:#f9d4d4, font-weight:bold, font-size:14px; classDef eeg fill:#d4f9d4, font-weight:bold, font-size:14px; classDef stimulation fill:#d4d4f9, font-weight:bold, font-size:14px; classDef motorimagery fill:#f9f9d4, font-weight:bold, font-size:14px; classDef patient fill:#f9d4f9, font-weight:bold, font-size:14px; classDef system fill:#d4f9f9, font-weight:bold, font-size:14px; A[Marc Sebastian] --> B[EEG cap, stimulators, software. 1] A --> C[16 EEG electrodes over motor cortex. 2] C --> D[Correctly placed cap, conductive gel. 3] A --> E[Stimulator electrodes over muscle belly. 4] E --> F[50Hz frequency, adjustable amplitude. 5] E --> G[Avoid finger flexion, reposition electrodes. 6] E --> H[No pain, find right threshold. 7] A --> I[8s motor imagery tasks, 2s cue. 8] I --> J[Explain imagery, imagine reaching overhead. 9] I --> K[Concentration critical, >50% success factor. 10] I --> L[System remotivates declining patients. 11] I --> M[Provide instructions every session. 12] A --> N[3 runs, 80 trials each. 13] N --> O[Calibration mode: feedback always. 14] N --> P[Training mode: feedback if done well. 14] N --> Q[240 trials, 45min sessions. 15] A --> R[Accuracy % after each session. 16] R --> S[Reactivates motor cortex, accuracy improves. 17] A --> T[Patient selection, ERD/ERS, CSP plots. 18] T --> U[16 EEG lines, electrode signal quality. 19] T --> V[Set frequency, pulse width, current. 20] T --> W[Select patient, adjust parameters, apply electrodes. 21] W --> X[Check signal: eyes closed, mouth movement. 22] A --> Y[Auditory/visual cues for left/right imagery. 23] Y --> Z[Calibration: avatar moves always. 24] Y --> AA[Training: avatar moves if imagery successful. 24] Y --> AB[ERD/ERS shows motor cortex activation. 25] Y --> AC[Therapist quiet to avoid EEG artifacts. 26] Y --> AD[Increase intensity if too weak. 27] Y --> AE[Imagine movement, let stimulation activate muscles. 28] Y --> AF[No concentration, no feedback provided. 29] A --> AG[Accuracy improves, classifier more precise. 30] class A,B recovery; class C,D,U,X,AB eeg; class E,F,G,H,V,AD stimulation; class I,J,K,L,M,Y,Z,AA,AC,AE,AF motorimagery; class N,O,P,Q,R,S,AG system; class T,W patient;

Resume:

1.- The recovery system includes an EEG cap, functional electrical stimulators, and software. The patient sits looking at a monitor displaying an avatar.

2.- 16 EEG electrodes are placed over the motor cortex. Functional electrical stimulators deliver stimulation to the forearms.

3.- The EEG cap must be placed correctly, not rotated. Electrodes require conductive gel, about 2-2.5mL total for the 16 electrodes.

4.- Placing the functional electrical stimulator electrodes can be tricky at first. They should be over the muscle belly to avoid pain.

5.- Stimulation is delivered at 50Hz frequency. Amplitude is increased slowly for each patient until a nice hand opening movement is seen.

6.- If electrodes are placed too medially, finger flexion rather than extension occurs. Electrodes may need to be repositioned.

7.- Patients should not feel pain during stimulation. Finding the right threshold can be tricky at first but gets easier with experience.

8.- Each motor imagery task lasts 8 seconds. After a 2 second cue, the patient imagines the movement for the remaining time.

9.- Explaining motor imagery to patients who lost limb function can be challenging. Imagining reaching for an object overhead can help.

10.- Patient concentration is critical, accounting for over 50% of therapy success. Feedback only occurs if motor imagery is performed.

11.- Patient motivation often declines over time in rehabilitation. The recovery system's technology can remotivate patients.

12.- Instructions should be provided every session, regardless of patient progress. Pause times can be used to remotivate the patient.

13.- Sessions include 3 runs of 80 trials each (40 left, 40 right). The first run calibrates the system to brain changes.

14.- In calibration mode, feedback is provided regardless of performance. Training mode only provides feedback if motor imagery is done well.

15.- 240 total motor imagery trials are performed per session (120 left, 120 right) in randomized order. Sessions last about 45 minutes.

16.- The system provides a motor imagery accuracy percentage after each session indicating how well the patient can control left vs right.

17.- This motor learning process teaches patients to reactivate the motor cortex. Accuracy is expected to improve over the 25 sessions.

18.- The software has a patient selection screen and main screen showing ERD/ERS plots, CSP plots, and real-time accuracy.

19.- 16 EEG lines are shown. Electrode signal quality is indicated by color. ERD/ERS maps show motor cortex activation.

20.- Parameters like frequency (50Hz), pulse width (300ms for upper, 400ms for lower limbs), and current are set for each patient.

21.- Setup involves selecting the patient, adjusting parameters, applying EEG cap electrodes in a standard order, and testing the EEG signal.

22.- Signal quality is checked by having the patient close eyes (alpha waves appear) and move mouth (artifacts seen).

23.- Once signal is stable, the session can begin. The system provides auditory and visual cues for left/right motor imagery.

24.- During calibration, the avatar moves regardless of performance. In training, avatar movement depends on successfully performing motor imagery.

25.- Analyzing ERD/ERS plots shows if the patient is activating the correct motor cortex during left/right imagery.

26.- The therapist should remain quiet and still during the session to avoid disturbing the patient and introducing EEG artifacts.

27.- If stimulation feels too weak, the intensity can be increased between runs.

28.- In training mode, the patient should concentrate on imagining the movement and let the stimulation activate their muscles.

29.- Lack of concentration results in no feedback being provided, as the system can detect if motor imagery is being performed.

30.- Motor imagery accuracy improves over sessions as the classifier becomes more precise with additional data.

Knowledge Vault built byDavid Vivancos 2024