CV

Teaching Assistant for ENGN 2520

Brown University, School of Engineering

I'm the teaching assistant for ENGN 2520 Pattern Recognition and Machine Learning, taught by Dr. Leo Kozachkov.

Graduate Research Assistant

BrainGate Team

I'm working on iBCI system to restore communication for ALS patients. And also investigating how we can utilize deep learning methods to facilitate iBCI fast recalibration.

Algorithm Intern

Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences

I was in charge of a clinical project to investigate the brain network connectivity changes during visual search and recognition using sEEG data.

Master of Science

Brown University

Biomedical Engineering

• Machine Learning
• Deep Learning
• Statistical Neuroscience
• Learning with Limited Labelled Data
• Neural Engineering
• Dynamical Systems
• Neural Dynamics

Bachelor of Engineering

Southern University of Science and Technoloty

Biomedical Engineering

Brown University Academic Sub-track Summer Research Stipend ($5,000.00)

Brown University Institute of Biology, Engineering, and Medicine

Brown University Institute of Biology, Engineering, and Medicine Master’s Scholarship ($8,567.00)

Brown University Institute of Biology, Engineering, and Medicine

SUSTech Freshman Scholarship, First Prize (¥30,000.00)

Southern University of Science and Technology

Short-Interval Priming Effects: An EEG Study of Action Observation on Motor Imagery

IEEE Transactions on Cognitive and Developmental Systems

How to effectively set and adjust training tasks is the main challenge of task-oriented rehabilitation. Recent studies have shown that motor priming effects could be used as an important reference for task setting. However, the use of assessment scales to quantify priming effects lacks timeliness. This study focused on evaluating the motor priming effects on a smaller timescale using electroencephalogram (EEG). Six healthy adults were recruited to perform motor imagery (MI) of elbow flexion extension primed by the same action observation (AO). The control task replaced AO with arrows indicating different elbows. The motor priming effects were evaluated by the event-related desynchronization (ERD) of μ band (8–13 Hz) power and single-trial classification of left or right elbow MI. Both ERD intensity and classification accuracy of MI primed by AO were more significant than MI instructed by arrows. The results demonstrated that short-interval motor priming effects could be evaluated at the neural level and could be used as feasible neurofeedback to guide task-oriented rehabilitation.

Characterization of Bimanual Cyclical Tasks From Single-Trial EEG-fNIRS Measurements

IEEE Transactions on Neural Systems and Rehabilitation Engineering

Robot-assisted bimanual training is promising to improve motor function and cortical reorganization for hemiparetic stroke patients. Closing the rehabilitation training loop with neurofeedback can help refine training protocols in time for better engagements and outcomes. However, due to the low signal-to-noise ratio (SNR) and non-stationary properties of neural signals, reliable characterization of bimanual training-induced neural activities from single-trial measurement is challenging. In this study, ten human participants were recruited conducting robot-assisted bimanual cyclical tasks (in-phase, 90° out-of-phase, and anti-phase) when concurrent electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) were recorded. A unified EEG-fNIRS bimodal signal processing framework was proposed to characterize neural activities induced by three types of bimanual cyclical tasks. In this framework, novel artifact removal methods were used to improve the SNR and the task-related component analysis (TRCA) was introduced to increase the reproducibility of EEG-fNIRS bimodal features. The optimized features were transformed into low-dimensional indicators to reliably characterize bimanual training-induced neural activation. The SVM classification results of three bimanual cyclical tasks revealed a good discrimination ability of EEG-fNIRS bimodal indicators (90.1%), which was higher than that using EEG (74.8%) or fNIRS (82.2%) alone, supporting the proposed method as a feasible technique to characterize neural activities during robot-assisted bimanual training.

Communication Restoration for an Advanced ALS Patient

Designed and implemented iBCI decoders for restoring communication in stroke and ALS patients, utilizing Machine Learning and Deep Learning techniques. The system has been the only communication method for more than 6 months. Enabled communication for an advanced ALS patient with a multiclass fast-deploy Gaussian Naïve Bayes iBCI decoder, reaching over 90% accuracy for 5-way online decoding. Developed a sleep detection method for locked-in ALS patients based on the LFP signal, reaching over 90% accuracy.

• Real-time Neural Signal Decoding
• Python
• Machine Learning

Meta-Learning for Stable Long-term iBCI Decoder

Improve decoding stability by minimizing the need for recalibration and enabling same-day adaptation, reducing the recalibration time to within 1 minute. Train and evaluate the model on the supercomputer OSCAR for higher running speed.

• Deep learning
• Meta-learning
• PyTorch
• Neural Signal Processing

Brain Connectivity Changes in Different Visual States

Investigated brain connectivity changes during visual search and recognition processes to better understand the top-down and bottom-up visual pathways and the mechanism of attention. Led the entire project, including building the virtual reality experiment in Unity, conducting clinical experiments, programming the sEEG data analysis pipeline, and performing spike sorting and analysis.

• Brain Network
• Granger Causality
• Clinical Recording
• sEEG