Pan Tony, Huang Kun
Department of Biomedical Informatics, The Ohio State University, Columbus, Ohio 43210. Email:
Conf Proc IEEE Eng Med Biol Soc. 2005;2005:3112-6. doi: 10.1109/IEMBS.2005.1617134.
Microscopic imaging is an important phenotyping tool to characterize the phenotype (e.g., morphology and behavior) change caused by genotype manipulation such as mutation and gene knockout. Recently we use high resolution microscopic imaging to study the morphological change on mouse placenta induced by retinoblast (Rb) gene knockout. In order to assess the morphological change we first segment each microscopic image into regions corresponding to different tissue types. Due to the complex structure of these tissues and large variation among the more than 2,000 images, we design a Bayesian supervised segmentation method which utilizing image features of all levels. The method has been applied to the entire data set and generated satisfactory results that is essential for further analysis on 3-D morphological change of the tissue types.
