Research Experience

Developed a data driven analysis method for predicting brain states based on visibility graph and machine learning algorithms. It was successfully applied on the dataset collected from six mice subjects using widefield calcium imaging and achieved significantly better classification performance compared to conventional approach. This work had been orally presented on the conference of OSA CLEO 2017 [Slides]. It shows great potential in applying for the brain-computer interface. This project was collaborated with Dr. David Margolis and Dr. Christian Lee at Department of Cell Biology and Neuroscience, Rutgers University. The journal manuscript of this study had been submitted and is under review. This study was supported by Siemens, NSF, NJCBIR and NIH. [Slides]

Applied a novel averaging technique based on dynamic time wrapping algorithm for analyzing brain activity recorded by fNIRS. Compared to conventional averaging technique, it solved the underestimation problem in detecting brain activation by compensating the non-linearly variable latencies in the recorded data, and hence, strengthened the detection power. It was extensively validated by synthetic and human-based real data. This study had been published on Journal of Biomedical Optics and presented in the conference of OSA Biomed 2014. [Slides] [Poster]

Proposed an analysis framework, combining wavelet transform coherence and multivariate permutation tests, for investigating dynamicity of spontaneous brain functional connectivity. This method was demonstrated via widefield calcium imaging. This study had been orally presented on conference SPIE Optics+Photonics 2017. [Slides]

Proposed sensor arrangement for EEG-fNIRS multi-modal experiment via simulation and experimental studies. This arrangement is proved as the optimized one for capturing brain activities originated from same cortical locations by different modalities, and hence improved the accuracy in quantitatively modeling brain functionalities. This study was presented in major scientific conferences (fNIRS Conference 2016 and OSA Biomed 2016) and has been received wide notice. This study was supported by Siemens Healthineers. The journal manuscript of this study is under preparing. [Poster]

Investigated dynamicity of brain functional connectivity based on fNIRS using wavelet and permutation testing. The results suggested that fNIRS could capture the trial-to-trial variability in the brain activity and the recorded time series should not be treated as stationary in the data analysis. This study was orally presented in the conference of IEEE SPMB 2014 and The Signal and Information Processing Seminar at Rutgers in 2015 [Slides].

Attended the establishment of Brain Imaging Laboratory, participated in selecting key instruments and supplies, proposed lab regulations and managed daily activities, kept lab running safely and efficiently, trained and oversaw new members.

Designed and conducted extensive brain imaging experiments (including resting-state and task-based) on human subjects using EEG/fNIRS techniques. The collected data have been successfully used by several researchers for different projects. I have full experience in performing brain imaging experiments on EEG instrument Brain Vision, fNIRS instruments TechEn CW6 system, Hitachi ETG-4000, and NIRx NIRScout.

Project: IARPA Knowledge Representation in Neural Systems – Researched multivariate time series analysis for decomposing human MEG data with the aim of extracting spatiotemporal features related to the processing of semantic information, implemented machine learning algorithms to decoding brain states associated with “what subjects are watching” from brain imaging. The results confirmed that MEG brain imaging contains information about generic mental representations.

Supervisor: Dr. Francisco Pereira and Dr. Bin Lou.