Abstract
Advances in imaging techniques have yield massive images into the biology. Along with the increase of dimension and data size of bioimages in the research field, a need for computer-aided image analysis becomes clear. However, software environments are not utilized enough for image analysis. This is because the versatility of purposes and the diversity of bioimages. With this situation,we have developed an adaptive classification system for bioimages, named “clustering-aided rapid training agent (CARTA)”. The CARTA is applicable to various bioimage classification that facilitates annotation and selection of features. The CARTA interactively collects information from experts and generates the customized classifier for the specified bioimages. In this review, approaches for development of bioimage classifiers are outlined as follows: clustering, rule-based classifier, supervised learning and active learning. The explanation of CARTA is followed as an example of active learning approach.
