An improved method for the automated enumeration of fluorescently labelled bacteria in human faeces

The study aimed to improve microscopy-based automated recognition of faecal bacterial cells labelled with 16S rRNA-targeted oligonucleotides and 4',6-diamidino-2-phenylindole (DAPI). Based on the software KS400 (Carl Zeiss Vision, Hallbergmoos, Germany), designed for automising microscopy-based image capture and image analysis, a routine was developed that affords the recognition of doubly stained bacteria and the rejection of artefacts. The accuracy of the automated enumeration was investigated by comparing the resulting data with those obtained by manual counting. The newly developed method was subsequently used to compare the total bacterial counts in human faecal samples using the domain specific probe Eub338 alone and a mixture of 5 domain-specific probes, respectively. Faecal samples from 90 healthy volunteers were analysed. The cell counts obtained with Eub338 were 10% lower than those obtained with the probe mixture. Since the cells detected with the probe mixture covered a wide range of signal intensities, a dynamic analysis routine was developed to effectively detect the whole range of bright to weak signals within the same image, while at the same time reliably rejecting artefacts.