My Research

One of my current projects is to study deep learning models, such as CNN, RNN, or Reinforcement Learning, to solve medical imaging problems.

An iris recognition approach based on the detection and matching of iris crypts. This new method, mimicing fingerprint identification, allows human to supervise the procedure and exercise judgement.

Tracking biomedical objects, like bacteria or cells, is a foundamental problem for quantitative analysis of dynamics. I developed a tracking algorithm based on Earth Mover's Distance. The effectiveness has been demonstrated in various real applications.

Quantifying the morphology and topology of 3D fibrin network is important to study thromus. I deviced a simply but effective algorithm for the segmentation of interlacing 3D tubular structures, as well as the identification and analysis of the network structure.