Fully automatic determination of soil bacterium numbers, cell volumes, and frequencies of dividing cells by confocal laser scanning microscopy and image analysis

We describe a fully automatic image analysis system capable of measuring cell numbers, volumes, lengths, and widths of bacteria in soil smears. The system also determines the number of cells in agglomerates and thus provides the frequency of dividing cells (FDC). Images are acquired from a confocal laser scanning microscope. The grey images are smoothed by convolution and by morphological erosion and dilation to remove noise. The background is equalized by flooding holes in the image and is then subtracted by two top hat transforms. Finally, the grey image is sharpened by delineation, and all particles above a fixed threshold are detected. The number of cells in each detected particle is determined by counting the number of local grey-level maxima in the particle. Thus, up to 1,500 cells in 10 fields of view in a soil smear are analyzed in 30 min without human intervention. Automatic counts of cell numbers and FDC were similar to visual counts in field samples. In microcosms, automatic measurements showed significant increases in cell numbers, FDC, mean cell volume, and length-to-width ratio after amendment of the soil. Volumes of fluorescent microspheres were measured with good approximation, but the absolute values obtained were strongly affected by the settings of the detector sensitivity. Independent measurements of bacterial cell numbers and volumes by image analysis and of cell carbon by a total organic carbon analyzer yielded an average specific carbon content of 200 fg of C (mu)m(sup-3), which indicates that our volume estimates are reasonable.