Welcome to Han Peng’s homepage!

What’s going on?

  • State-of-the-art performance for brain age prediction! We won PAC 2019 brain age prediction challenge! The challenge consists of two goals: 1) Predict age from MRI brain images as accurate as posisble, and 2) Achieve accurate prediction results while keeping the age delta unbiased from age. We lead on both parts!
  • Deep learning with medical imaging data. I am now working on UK Biobank data to find patterns of disease and aging in the population. This is a postdoctoral project with Prof Steve Smith from FMRIB and Prof Andrea Vedaldi from VGG.
  • Neural network in spectrum image. I was working on using convolutional neural network to enhance the spectrum image features. Using CNN in scientific data can be tricky. The question is not only whether it works, but also how it works. Update coming soon (hope so :) …
  • Neuron image registration. I worked on some beautiful neuronal image data with Prof Kenneth Harris, who led me into the fantastic field of neuroscience. The goal in this project is to register an ex-vivo microscopic 2D image slice in the in-vivo 3D two-photon volume data so that we can link the information from different experiment methods. The data is large and can be noisy. Yet we find a method to register the different data sets with good accuracy and fast speed. Check out the GitHub site and the project page. Read more …
  • Graphene with a twist. Stacking 2D materials together is just like playing microscopic Lego. In the bilayer graphene scenario, the twist angle between the two layers is the driving parameter for the photonic device application and photochemical reactivity (Yin et al. 2016 and Liao et al. 2015). Peng et al. 2017 explains the underlying mechanism by using photoemission spectroscopy to visualize the electronic band structure’s evolution with the twist angle. I will give a talk in APS March meeting 2018. Read more…