|
My research lab at Janelia Farm Research Campus, Howard Hughes Medical Institute focuses on understanding the structures and functioning mechanisms of biosystems using data mining techniques for bioinformatics and computational molecular biology, particularly, bioimage data mining and informatics. These bioimages include cellular and molecular images and related medical images. Several recent projects are listed below. Some further description of my ongoing research can be found here.
- Novel Algorithms for 3D/4D/5D image/video computing, including image registration, neuron/fibrous-structure tracing, cell recognition and targeting, comparison and mining of complex patterns (e.g. neuron morphology), etc. We have developed a series of methods such as Reliable Landmark Matching (RLM) and Principal Skeleton (PS) for automated image alignment, Graph-augmented Deformable model (GD) and All-Path Pruning (APP) for neuron reconstruction, Simultaneous Recognition and Segmentation (SRS) of cells for image segmentation and recognition, etc.
- 3D Digital Atlas Modeling of Animals at the Single Cell/Neuron Level: building a 3D digital Worm Atlas for single cells in C. elegans and a 3D digital Fly Brain Atlas for fruitfly. Particularly, for the first time we have quantitatively assessed the spatial invariability of patterns in these systems, which are critical for using these atlases for new biological experiments.
- Analysis of Gene Expression Pattern Images: high-performance image analysis and mining for different model organisms, such as fruitfly, C. elegans, and mouse;
- Feature/Model Learning: developing algorithms and software to extract/select features and learn models from biological data (e.g. microarray data, neuronal gene expression patterns in microscopical images, etc.), such as the mRMR algorithm (minimum Redundancy Maximum Relavance) for feature selection.
- Brain Morphometry and Neuronal Structure Analysis: developing computational tools to measure and analyze quantities of brain tissues and neurons.
- High Performance Visualization and Analysis Tools for Large-Scale Bioimages and Their Databases: developing many tools that make it much easier to explore large scale data; particularly, we developed the Vaa3D platform, which we call a Swiss Army Knife for 3D/4D/5D image visualization and analysis.
I have also worked in the more traditional pattern recognition and artificial intelligence field. I do have continuous interest in many theoretical problems there. |
