PENGHAO
profile photo

Penghao Qian

In fact, humans did not discover the neuron; they reconstructed it.

— A Brief History of Simulation Neuroscience

I am currently a PhD student majored in Artificial Intelligence at School of Computer Science, Fudan University, Shanghai, where my supervisor is Prof. Hanchuan Peng, focusing the analysis of structural and functional brain networks at the single-cell level.
I got my Master degree of Computer Science at Institute for Brain and Intelligence, Southeast University, Nanjing, also under supervision of Prof. Peng, while working directly with the senior researcher Dr. Linus Manubens-Gil. I also worked closely with A.P. Lijuan Liu.
Before that, I worked as a R.A. for A.P. Dan Zhang at Tsinghua University and did my bachelors at the college of information and electrical engineering under the supervision of A.P. Xiang Li, China Agricultural University.

Research Areas: Computational Neuroscience, Brain Networks, Simulations, Artificial Intelligence, Signal Processing

News

Education and Internship

Education

Fudan University (FDU), Shanghai, China
2024.09 - Now
Southeast University (SEU), Nanjing, China
2021.09 - 2024.06
China Agricultural University (CAU), Beijing, China
2016.09 - 2021.06

Internship

Trainee
Shanghai Artificial Intelligence Laboratory, Shanghai

Supervisor: Dr. Zixin Liu
2024.12 - 2025.09
Research Assistant
Institute for Brain and Intelligence, Southeast University

Supervisor: Prof. Hanchuan Peng
2024.06 - 2024.09

Research Assistant
Department of Psychology, Tsinghua University

Supervisor: A.P. Dan Zhang
2017.09 - 2020.06

Projects

Unimportant, there's a million things I haven't done, just you wait, just you wait …

— Alexander·Hamilton

My study of neuroscience was a gradual deepening process. I started with the implementation of brain-computer interfaces to understand neural mechanisms. Then I started inter-individual analysis from physiological signals (e.g. EEG, sEMG, PPG). At the master's level, I started to study the structural and functional networks of the brain at the single cell level of neurons and developed novel morphology-based classification tools. After that I hope to carry out some research on more abstract models at the whole brain scale.

Analysis of structural and functional brain networks

In this project we divided into two parts. In the first part we studied the impart of neuronal morphology details on the network structure, especially the bouton distribution. In the second part, we studied the correlation between single-cell connectivity and function by simulations.

From single neuron to structure
Supervisor: Prof. Hanchuan Peng & Dr. Linus Manubens-Gil
2023.09 - 2024.06

Paper | Project Page

We examined the distribution of pre-synaptic contacts in axons of mouse neurons and constructed whole-brain single-cell neuronal networks using an extensive dataset of 1891 fully reconstructed neurons. We found that bouton locations were not homogeneous throughout the axon and also among brain regions. As our algorithm was able to generate whole-brain single-cell connectivity matrices from full morphology reconstruction datasets, we further found that non-homogeneous bouton locations have a significant impact on network wiring, including degree distribution, triad census and community structure. By perturbing neuronal morphology, we further explored the link between anatomical details and network topology. In our in silico exploration, we found that dendritic and axonal tree span would have the greatest impact on network wiring, followed by synaptic contact deletion. Our results suggest that neuroanatomical details must be carefully addressed in studies of whole brain networks at the single cell level.
From single neuron to function
Supervisor: Prof. Hanchuan Peng & Dr. Linus Manubens-Gil
2023.09 - Present

Paper

We simulated the whole mouse brain's resting state using an extensive dataset of 1876 fully reconstructed neurons, revealing stronger and more varied connections than previous tracer injection-based brain connectivity measurements indicated. After optimizing global coupling and background noise parameters, we tested the simulation's alignment with experimental data, finding that simulations using single-cell connectivity have increased predictive power compared to tracer-based connectomes. Our findings underscore the importance of incorporating detailed single-cell information to accurately model brain dynamics, offering insights into the mouse brain's functional architecture.

Tools for neuron classification based on manifold patterns

Supervisor: Prof. Hanchuan Peng & A.P. Lijuan Liu.
2022.05 - 2024.06

Paper | Morphological feature space toolkit | Preprint | Neuron type classification toolkit

As morphological feature spaces are often too complicated to classify neurons, we introduce a method to detect the optimal subspace of features so that neurons can be well clustered. We have applied this method to one of the largest curated databases of morphological reconstructions that contains more than 9,400 mouse neurons of 19 cell types. Our method is able to detect the distinctive feature subspaces for each cell type. Our approach also outperforms prevailing cell typing approaches in terms of its ability to identify key morphological indicators for each neuron type and separate super-classes of these neuron types. Subclasses of neuronal types could supply information for brain connectivity and modeling, also promote other analysis including feature spaces.

Study of EEG signal correlation between students

Supervisor: A.P. Dan Zhang
2018.10 - 2020.6

Paper | Code

This study examined the relationship between correlation of EEG signals and academic performance of students in the classroom. My work in it focused on the analysis and processing of the data.
In this study, a portable headband with two electrodes was used to collect real data from the classroom for up to four months. Therefore the data was large and there were also a large number of artifacts. For this I provided a process that involves slicing the data into 30s epochs and evaluating the data quality. Then, the slow drifts were removed using NoiseTool under Matlab and the ocular artifacts were attenuated with the MSDL (multi-scale dictionary learning) toolbox. At the end, calculation of total interdependence were also provided.

Game based on BCI and VR

Supervisor: A.P. Xiang Li
2017.9 - 2019.5

Project Page

Here is a project that tries to combine BCI and VR (Both EEG-BCI and VR require a headset. It is natural to hope that the two can be combined).
For BCI interaction, we think that active BCI, which requires external stimuli to induce interaction, is not natural enough, and passive BCI has been lacking typical interaction scenes, which cannot be replaced by other interaction methods, such as the measurement of "personal state" and "emotion". Therefore, we tried to design a game that combines various BCI interaction ways with personal state as an operation method.

Research

Only the written can stably exist outside the mind.

— Elowen·Crawford

I focus on studying brain structure and function from a network perspective at different scales (inter-individual, single-cell level), while developing new tools for classifying neuronal morphological categories.

Journal Papers:

Main works
  1. Non-homogenous axonal bouton distribution in whole-brain single cell neuronal networks
    Penghao Qian, Linus Manubens-Gil*, Shengdian Jiang, Hanchuan Peng*
    Cell Reports | 2024 | Paper | Code
  2. Cell Typing and Sub-typing Based on Detecting Characteristic Subspaces of Morphological Features Derived from Neuron Images
    Sujun Zhao, Penghao Qian, Lijuan Liu*
    Preprint | 2023 | under review | Preprint | Code
  3. Inter-brain Coupling Reflects Disciplinary Differences in Real-world Classroom Learning
    Jingjing Chen, Penghao Qian, Xinqiao Gao, Baosong Li, Yu Zhang*, Dan Zhang*
    npj Science of Learning | 2023 | Paper | Code
  4. Manifold-Classification of Neuron Types from Microscopic Images
    Lijuan Liu*, Penghao Qian
    Bioinformatics | 2022 | Paper | Code
Participation
  1. Neuronal diversity and stereotypy at multiple scales through whole brain morphometry
    Yufeng Liu, Shengdian Jiang, Yingxin Li, ... , Penghao Qian, ... , Hanchuan Peng*
    Nature Communications | 2024 | Paper | Project

Conference:

  1. BioImage Informatics Conference | Bioimaging and microscopy applications | Poster
    Poster Section | Institut Pasteur Online 2021
    Title: Single neuron morphological details imply a shift from a Small-World to a Scale-Free topology in the mouse brain network
    Penghao Qian*, Linus Manubens-Gil
  2. 3rd Annual Conference on Engineering Psychology of C.P.S. | News
    Assist with Oral Presentation | East China Normal University 2019
    Our analysis of EEG data collected by portable devices were presented by A.P. Dan Zhang

Summer School:

  1. The Computational and Cognitive Neuroscience (CCN) summer school
    Trainees | Cold Spring Arbor Asia 2024
  2. BioBit Program Summer School for Computational Biology | Poster
    Best Poster and Best Student Award | Zhejiang lab 2023
  3. IEEE 4th International Summer School for Neural Engineering | News
    Trainees (40 selected from 400+ applicants) | Tsinghua University 2018

Competition:

  1. Modeling of deep brain electrical stimulation (DBS) therapy for Parkinson's disease | Project
    Won the National 2nd Prize | China Postgraduate Mathematical Contest in Modeling | 2021
    • Construct neuron firing basic model based on the H-H model and the structure of basal ganglion network.
    • Compare the firing of ganglion circuits in Parkinson's disease with normal conditions.
    • Operate DBS in two different target (STN and GPi) and select the optimal stimulation target.
    • Change the intensity, frequency and stimulation mode of DBS to get the optimal parameters combination.
    • Also find other optimal stimulation targets.
  2. Design of Dynamic Scheduling Strategy of Smart Rail Guided Vehicle (RGV)
    Won the National 2nd Prize | Contemporary Undergraduate Mathematical Contest in Modeling | 2018
    • Simulated the RGV operation process with different algorithm (Sequences, Elevator Scheduling, Greedy).
    • Prune the feature search space to find the optimal solution.
    • Discuss all cases of initialization and the influence on the subsequent processes.
    • Estimate process arrangement and failure risk to prove the system's good robustness.

Services

Honors

Honorable Title

Awards

Scholarship

Last updated:

Welcome to use this website's source code, just add a link back to here.
No. Visitor Since Jun 2020. Powered by w3.css