Quantification and sizing of filamentous cyanobacteria in environmental samples or cultures are timeconsumingand are often performed by using manual or semiautomated microscopic analysis. Automation ofconventional image analysis is difficult because filaments may exhibit great variations in length and patchyautofluorescence. Moreover, individual filaments frequently cross each other in microscopic preparations, asdeduced by modeling. This paper describes a novel approach based on object-oriented image analysis tosimultaneously determine (i) filament number, (ii) individual filament lengths, and (iii) the cumulativefilament length of unbranched cyanobacterial morphotypes in fluorescent microscope images in a fully automatedhigh-throughput manner. Special emphasis was placed on correct detection of overlapping objects byimage analysis and on appropriate coverage of filament length distribution by using large composite images.The method was validated with a data set for Planktothrix rubescens from field samples and was compared withmanual filament tracing, the line intercept method, and the Utermohl counting approach. The computerprogram described allows batch processing of large images from any appropriate source and annotation ofdetected filaments. It requires no user interaction, is available free, and thus might be a useful tool for basicresearch and drinking water quality control.
展开▼
