|
(43-08) Partial automatic image analysis for identification and quantification of phytoplankton in ecotoxicology. Hense, Burkhard*,1, Rodenacker, Karsten1, Juetting, Uta1, Gais, Peter2, 1 GSF - National Research Center for Environment and Health, Neuherberg, Germany2 GSF - National Research Center for Environment and Health, Neuherberg, Germany ABSTRACT- A high number of studies deal with effects of stressors to aquatic ecosystem. For a full protection it is often desirable to get information on as detailed taxonomic level as possible. However, identification and quantification e.g. of algae is a time consuming work to be done by a qualified person and thus expensive. Therefore many studies do not go into taxonomic detail, but use sum parameters like concentration of chlorophyll a or accessory pigments. Here a method is presented, using automated image acquisition, evaluation and recognition to ease the time part of manual organism recognition and counting at the microscope. Phytoplankton samples prepared in plankton chambers following the Utermoehl method were used. A newly developed automated software system enables to acquire images applicable for image analytical and statistical evaluation with an automated inverse microscope. Interactivity allows to adapt the acquisition to the specific research demands and the design of training sets for new taxa. The subsequent image analysis consists of several steps, including rejection of unsuitable images, segmentation of structures, and classification. As far as the system is developed until now, every step of the whole process has considerably reduced the amount of human interaction time. Final objective is to get a semi-automated system of algae population profiling adequate for biologist. Key words: algae toxicity, community, image analysis, phytoplankton |
