Automated measurements of filamentous cyanobacteria by digital image analysis

Abundance and biomass estimations of filamentous cyanobacteria often use conventional microscope measurements, which are highly time- consuming and provide no information on filament morphology. In this article, we describe an image analysis method developed for rapid, semiautomated or fully automated estimations of filament lengths and abundance of uncoiled cyanobacteria. The method was developed for the Baltic Sea cyanobacterium Aphanizomenon sp. and then applied on Trichodesmium sp. strain IMS101 to demonstrate the wider applicability of the system. Digitized microscope images were analyzed with software that identified the filaments by comparing their dimensions with specified acceptance ranges and automatically measured selected parameters. Validation of the method against estimates obtained by conventional microscope measurements showed excellent agreement for both Aphanizomenon sp. and Trichodesmium sp. Compared with previously described image analysis methods based on fluorescence, the presented method is an improvement, as filament dimensions are used to discriminate between the cyanobacterial filaments and other objects. The image analysis also provides detailed morphological information on individual filaments, and the fully automated procedure requires only computer time for counting and is therefore highly cost- efficient.