Pattern Recognition of Bacteria Using Scanning Electron Microscopy and Materials Analysis.

A need for rapid characterization of bacteria for prompt clinical diagnosis and early warning of unusual bacteria in the atmosphere has prompted a number of different approaches to attempt to circumvent the usual time-consuming tedious procedures commonly employed. None appear to have been very successful. In the present case, state-of-the-art scanning electron microscopy SEM, energy dispersive x-ray analysis EDXA and automated image analysis by computer were employed in a survey of thirteen species of bacteria. The organisms were first cultured and cloned to check for mutants or other contaminants and were then fixed, stained and dispersed, free of debris on a structure-free background. Next, the magnification and other SEM operating conditions were controlled, and the binary image in the secondary electron mode was made to correspond to the grey scale image. Finally, the computer printout of each individual organism was checked with the binary image to pick up inconsistencies in the data arising from the characteristics of the stereologic based program. The computer data for the mean width and average length were plotted on a ternary diagram. After the protocol was established and the variables controlled in each sample, about 100 organisms were measured and then the entire procedure was duplicated to determine reproducibility. In three runs on thirteen organisms, the reproducibility was generally around 1, easily sufficient to characterize each organism. This preliminary study indicates a feasibility that the enterics and other bacteria of interest to man could be rapidly identified by their genetically characteristic morphology.