抄録
Recent advances in imaging equipment have enabled the acquisition of many kinds of bioimages in huge numbers. With the acquisition of such imagery, computer assistance becomes increasingly important for image inspection. To provide an automated and versatile bioimage classification system, we have developed an active learning algorithm combined with a genetic algorithm and self-organizing map named Clustering-Aided Rapid Training Agent (CARTA). Using CARTA, similar images can be drawn from many images. Applying this feature of CARTA, we are developing a framework for the detection of similar cellular architectures in wide-field fluorescence microscopic images. In this article, we describe an example case of semi-automatic detection of stomatal regions from a fluorescence microscopic image of Arabidopsis leaf surface cell contours.
