About me
My journey in neuroscience started with “How the retina works”, a delicate, complex tissue that serves as the gateway to vision. I completed my PhD (2011–2016) at Peking University Health Science Center under the supervision of Prof. Mingliang Pu, focusing on the fundamental mechanisms of how the retina works.
Driven by a need to decode the structural “ground truth” of neural organization, I transitioned into the field of connectomics during my postdoctoral fellowship with Prof. Tiejun Huang at the School of Computer Science, Peking University. Following this, I spent two years at the Life Simulation Center of BAAI, reporting to Dr. Lei Ma, where I led efforts to integrate and open-source pipelines for large-scale connectomic data processing. In 2023, I joined the labs of Prof. Heping Cheng and Dr. Lei Ma at the National Biomedical Imaging Center (NBIC) to develop more efficient and robust methods for Bio-EM data.
Currently, I serve as an engineer at Platform IV of the Multi-modal cross-scale Biomedical Imaging Facility Flatform IV, specializing in high-performance, general-purpose image processing solutions on large-scale clusters.
The central question driving my work is: How do animal eyes—specifically the retina—sample visual scenes and encode them into neural information? To address this, my research focuses on three key pillars:
- Retinal Sampling Mechanisms
- Every animal species has its own distinct retinal cell distribution pattern, what the reason and function is? Some quanlitative descriptions of this exist, but quantitative consideration is still lacking.
- A major difference between animal eyes from cameras is the non-uniform sampling, because perceiving rapidly is much more important than seeing clearly to animal survival. How can this rule inspire artificial designs?
- Multimodal Retinal Modeling
- Retinal neuronal connections are ordered but complex. Connectomics can provide more comprehensive and precise observations.
- Physiological recording of retinal neuronal activity through microscopy can provide wider observation fields and thus many novel insights.
- Efficient bio-image processing
- Above researching techniques development promote data (image here) production, efficient processing pipelines are essential to converting images to knowledge (I2K).
- I’d rather to take image processing as the “retina” of computational vision, which deals with raw data for later better understanding.