Filtering and Registration of Cell and Intracellular Images

Recent advances in imaging technology demands to develop advanced image processing tools in modern science and engineering based on real-world data. Especially, quantitative image analysis has become crucial to molecular cell biology because of rapid advances of confocal laser scanning microscopes and fluorescent staining methods. Thus, simulations and analysis based on the observed images will be popular soon in systems biology and bioinfomatics due to rising of bioimage informatics. Although image processing and its related fields such as computer graphics, computer vision, and pattern recognition have been the most intensively studied fields in computer science, their applicability and effectiveness for such recent cell and intracellular images are not fully investigated.

In this talk, I will present my recent work on two important topics of biomedical image processing; filtering and registration. Since cell and intracellular images are usually corrupted with a lot of noise, it is important to filtering out noise on the images for further analysis. In first part of the talk, I will introduce a novel edge-aware filtering method based combining the non-local means, evolutionary PDE, and the Taubin's methods in order to produce high quality denoising results. Image registration is a powerful tool in many applications such as medical image analysis, morphing in CG, shape reconstruction in computer vision and CAD, and so on. Registration on volumetric cell and intracellular images has also an considerable attention to compare separately observed images e.g. gene expression analysis, morphological analysis of cell cytoskeletons, and quantitative comparisons for organelles. In second part of the talk, I will present a novel framework for intracellular volume registration based on extending the standard registration framework in 2D image processing to curved cell geometry and cell topology. The talk includes theoretical backgrounds of the methods and the numerical experiments on real-world cell and intracellular images.