Computer Assisted Microscopy *

A major challenge in microbial ecology is to develop reliable and facile methods of computer-assisted microscopy that can analyze digital images of complex microbial communities at single cell resolution and compute useful ecological characteristics of their organization and structure. Current systems are limited in that they can only perform object classification of up to three bacterial morphotypes: straight rods, spheres, and curved rods. The first version of CMEIAS(c) (Center for Microbial Ecology Image Analysis System, developed at MSU) can automatically classify each bacterial cell into one of 11 morphotypes (regular straight rods, cocci, spirals, curved rods, U-shaped rods, unbranched filaments, ellipsoids, clubs, prosthecate rods, rudimentary branched rods, and branched filaments) at an overall classification accuracy of 97%.

The second version of CMEIAS(c) (currently under development) has efficient semi-automated segmentation tools to reduce the microbial community image to objects of interest. It incorporates various new measurement features to compute the degree of morphological diversity, microbial abundance, and various spatial distribution relations of the microbes. The color recognition capability of CMEIAS(c) is being developed to facilitate the use of fluorescent molecular probes.

The project (and one of the goals for CMEIAS(c)-3) is to analyze a database of cell size range distributions to select the appropriate borders for the size classification of each of the 11 morphotypes that are currently identifiable. For example, in a given microbial community, a scatter plot of the cell length verses cell width for the resident regular straight rods will likely show several clusters of points, suggestive of several distinct operational morphological units (OMUs) for this morphotype. It is desired to identify the appropriate size border for the classification scheme that distinguishes these regular straight rods as different OMUs based upon their length and width. Similarly, it is desired to construct a classification scheme based upon appropriate parameters (length, width, area, radius of curvature, branching complexity, etc.) for each of the other ten morphotypes.

The ideally completed project deliverable would be a classification scheme, a database, and an interactive Excel macro package that can be added to CMEIAS(c)-3 to assist operators in the identification of appropriate size borders for the ``user-defined'' size and shape morphotype classifier of bacteria.

* This summary prepared by R. E. Svetic with the assistance of F. B. Dazzo, Professor of Microbiology at Michigan State University.

Back To the Top