Related Experiment Video
Updated: May 24, 2025
09:55
Monitoring Acupuncture Effects on Human Brain by fMRI
Published on: April 8, 2010
15.3K
Neural Manifold Decoder for Acupuncture Stimulations With Representation Learning: An Acupuncture-Brain Interface
IEEE Journal of Biomedical and Health Informatics
|March 3, 2025
Summary
This study decodes acupuncture manipulations using deep learning on EEG data, linking somatosensory stimulation to brain activity. The findings reveal distinct neural dynamics for different needling techniques, enhancing understanding of acupuncture
Area of Science:
- Neuroscience
- Biomedical Engineering
- Traditional Chinese Medicine
Background:
- Acupuncture's modulation of brain activity and cognitive function is known, but the precise link between somatosensory stimulation and dynamic brain responses remains unclear.
- Understanding this correlation is crucial for elucidating the clinical efficacy of acupuncture, particularly in neurological disorders.
Purpose of the Study:
- To propose a deep learning framework for decoding acupuncture manipulations using electroencephalographic (EEG) activity.
- To establish a correlation between specific acupuncture techniques (lifting-thrusting, twisting-rotating) and distinct brain activity patterns.
- To develop an acupuncture-brain interface for linking somatosensory input with neural representations.
Main Methods:
- Utilized contrastive representation learning and domain adaptation to analyze video recordings of acupuncturists, inferring 3D hand postures and motion trajectories.
- Applied unsupervised manifold learning to estimate low-dimensional latent neural manifolds from EEG signals evoked by acupuncture.
- Designed a nonlinear neural network decoder to transform neural manifolds into behavior manifolds, predicting manipulation and needling processes.
Main Results:
- Observed distinct transition dynamics in behavior manifolds for different acupuncture manipulations (lifting-thrusting vs. twisting-rotating).
- Successfully estimated latent neural manifolds from EEG signals that reliably represent acupuncture stimulations.
- Achieved high performance in decoding four types of acupuncture manipulations with 92.42% precision using the proposed EEG decoder.
Conclusions:
- The developed deep learning framework effectively decodes acupuncture manipulations from EEG signals, establishing a link between somatosensory stimulation and brain responses.
- The proposed acupuncture-brain interface provides a novel method for investigating the neural mechanisms underlying acupuncture's therapeutic effects.
- This approach offers an effective scheme for revealing the clinical efficacy of acupuncture treatment by quantifying brain activity changes.
