Knowledge Vault 3/16 - G.TEC BCI & Neurotechnology Spring School 2024 - Day 2
Exploring the neural mechanisms of normal and abnormal brain functions in humans using BCI
Shenghong He, Oxford University (UK)
<Resume Image >

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

graph LR classDef phd fill:#f9d4d4,stroke-width:2px classDef bci fill:#d4f9d4,stroke-width:2px classDef dbs fill:#d4d4f9,stroke-width:2px classDef parkinsons fill:#f9f9d4,stroke-width:2px classDef tremor fill:#f9d4f9,stroke-width:2px A[ Shenghong He,] --> B[PhD: brain-computer
interfaces, China. 1] A --> C[BCIs: record, process,
control signals. 2] C --> D[Non-invasive BCIs:
EEG, EOG, PhD. 3] C --> E[Oxford: invasive BCIs,
movement disorders. 4] C --> F[Projects: BCIs, disorders
of consciousness. 5] A --> G[Parkinson's: motor, cognitive
symptoms, treatments. 6] G --> H[DBS: subthalamic nucleus,
side effects. 7] G --> I[Beta oscillations: correlate
symptoms, DBS. 8] I --> J[Beta bursts: pathological,
reduced by treatment. 9] I --> K[Neurofeedback: suppress beta,
basketball game. 10] K --> L[Neurofeedback: reduces beta,
improves reaction. 11] I --> M[Adaptive DBS: beta-triggered,
more effective. 12] I --> N[Movement: beta suppressed,
gamma increased. 13] I --> O[Gamma: reduces in
continuous DBS. 14] I --> P[Beta: modulated by uncertainty,
reaction times. 15] I --> Q[TMS: beta phase affects
motor excitability. 16] J --> R[Neurofeedback: causally test
beta-motor relationship. 17] A --> S[Essential tremor: DBS
treatment, side effects. 18] S --> T[Closed-loop DBS: decodes
tremor-provoking movement. 19] S --> U[Adaptive DBS: reduces
tremor, less energy. 20] U --> V[Movement detection: machine
learning, ECoG, LFP. 21] S --> W[DBS: delivered at
specific tremor phase. 22] S --> X[Thalamic connectivity: predicts
DBS effect. 23] S --> Y[DBS targets:
VIM, motor cortex. 24] G --> Z[DBS: less effective
for Parkinson's gait. 25] Z --> AA[PPN: potential DBS
target for gait. 26] AA --> AB[PPN: alternating activity,
aligns with gait. 27] AA --> AC[Gait phase decoding:
optimize DBS timing. 28] A --> AD[Machine learning: extract
precise DBS signals. 29] A --> AE[Wireless systems: enable
closed-loop DBS optimization. 30] class A,B phd; class C,D,E,F bci; class G,H,I,J,K,L,M,N,O,P,Q,R,Z,AA,AB,AC,AD,AE dbs; class G,H,I,J,K,L,M,N,O,P,Q,R,Z parkinsons; class S,T,U,V,W,X,Y tremor;


1.-Sheng Hong He did his PhD on brain-computer interfaces in China and is now a postdoc at Oxford University.

2.-Brain-computer interfaces record brain signals, process them, and use control signals for various applications like communication, control, and understanding diseases.

3.-He studied non-invasive BCIs using EEG and EOG in healthy people and those with spinal cord injury during his PhD.

4.-At Oxford, he uses invasive BCIs with local field potentials from deep brain stimulation to treat movement disorders like Parkinson's.

5.-He's also involved in projects using BCIs for people with disorders of consciousness.

6.-Parkinson's disease has motor and cognitive symptoms. Levodopa medication helps early on, but deep brain stimulation is used in advanced stages.

7.- DBS in the subthalamic nucleus dramatically improves motor symptoms, but has side effects and may require stimulation adjustments over time.

8.-Beta oscillations (13-30 Hz) in the STN correlate with medication status and symptom improvement. DBS suppresses this beta activity.

9.-Longer beta bursts (>500 ms) are more pathological and correlate with Parkinson's symptoms. Medication and DBS reduce beta burst duration.

10.-He designed a neurofeedback protocol to train Parkinson's patients to suppress beta bursts using a basketball computer game.

11.-Healthy participants and Parkinson's patients could reduce beta activity and improve reaction times after the neurofeedback training.

12.-An adaptive DBS system triggered by real-time STN beta activity was more effective than continuous DBS while delivering less stimulation energy.

13.-During movement, STN beta is suppressed but gamma increases. Adaptive DBS was equally effective as continuous DBS during movement.

14.-Gamma activity also reduces more during continuous vs adaptive DBS. There may be an optimal balance between beta and gamma suppression.

15.-STN beta oscillations are also modulated by uncertainty and correlated with slower movement reaction times under high uncertainty.

16.-TMS delivered at specific beta phases had differing effects on movement, suggesting beta phase impacts motor cortical excitability.

17.-He proposed training Parkinson's patients to suppress beta bursts using neurofeedback BCI to causally test the beta-motor relationship.

18.-Essential tremor causes tremor during posture holding or movement, but not at rest. DBS can treat it but has side effects.

19.-A closed-loop DBS system for essential tremor was developed to decode tremor-provoking movement and only stimulate during those periods.

20.-This adaptive DBS for essential tremor reduced tremor as effectively as continuous DBS while delivering 60% less stimulation energy.

21.-The adaptive DBS system detects movement using machine learning on cortical ECoG or LFP signals and switches stimulation on/off accordingly.

22.-In another approach, DBS is delivered at a specific tremor phase for maximal suppressive effect at lower stimulation intensities.

23.-Connectivity between the thalamus and tremor, especially in the contralateral thalamus, predicts DBS effect better than cortex-tremor connectivity.

24.-Patients with stronger thalamic connectivity may benefit more from VIM DBS, while others may need alternate targets like the motor cortex.

25.-DBS is less effective for gait issues in Parkinson's. An alternating DBS pattern aligned with gait rhythm improved gait.

26.-The pedunculopontine nucleus (PPN) may be a better DBS target for gait as it shows increased activity during stepping movements.

27.-PPN local field potentials show alternating activity patterns aligned with gait phase, which could inform closed-loop alternating DBS.

28.-A real-time gait phase decoding embedded DBS system could optimize stimulation timing to the locomotor rhythm to improve gait.

29.-While current adaptive DBS for Parkinson's uses simple thresholding, machine learning could extract more precise stimulation control signals.

30.-Testing advanced closed-loop DBS algorithms is limited by short time windows after implantation, but wireless systems could enable better optimization.

Knowledge Vault built byDavid Vivancos 2024