BioImage Informatics: the challenge of knowledge extraction from biological images

Bioimage Informatics is a rapidly growing research field that is giving fundamental contributions to research in biology and biomedicine aiming at facilitating the extraction of quantitative information from images. Great advances in biological tissue labeling and microscopic imaging are radically changing how biologists visualize and study the molecular and cellular structures. These devices nowadays produce terabyte-sized multi-dimensional images: how to automatically and efficiently extract objective knowledge from such images has become a major challenge. In this manuscript we analyze the state-of-the-art of Bioimage Informatics, with a special focus on neuroscience. We show that there are increasing efforts to deliver methods and software tools providing functionalities for visualization, representation, management and analysis of 3D multichannel images. Nevertheless, most of them have been applied on datasets with size of MVoxel or few GVoxel, where the variations in contrast, illumination, as well as object shape and dimensions are limited. The huge dimensions of new 3D image stacks therefore ask for fully automated processing methods, whose parameters should be dynamically adapted to different regions in the volume. In this respect, this manuscript deepens in a recent contribution that digitally charts the Purkinje cells of whole mouse cerebellum, corresponding to an image dataset of 120 GVoxels.