Automatische klassifikation von bildzeitreihen für toxikologische hochdurchsatz-untersuchungen

Abstract

In this paper we present a new image-based method for automatic classification of spontaneous movement and heartbeat in toxicologica treated zebrafish. By combining new image acquiring strategies with image processing and feature extraction out of image sequences a set of classifiers is trained. Subsequently, these classifiers are iteratively combined and applied to three compounds to prove the feasibility in a prototype assay. © Oldenbourg Wissenschaftsverlag.