Morphotypic analysis and classification of bacteria and bacterial colonies using laser light-scattering, pattern recognition, and machine-learning system

Light scattering is one of the most fundamental optical processes whereby electromagnetic waves are forced to deviate from a straight trajectory by non-uniformities in the medium that they traverse. This presentation summarizes our recent research on application of light-scatter measurements paired with machine learning and pattern recognition methodologies for label-free classification of bioparticles. Two separate examples of light scatter-based techniques are discussed: forward-scatter measurements of bacterial colonies in an imaging system, and flow cytometry measurements of scatter signals formed by individual bacterial particles. Recently, we have reported a first practical implementation of a system capable of label-free classification and recognition of pathogenic species of Listeria , Salmonella, Vibrio, Staphylococcus, and E. coli using forward-scatter patterns produced by bacterial colonies irradiated with laser light. Individual bacteria in flow also form complex patterns dependent on particle size, shape, refraction index, density, and morphology. Although commercial flow cytometers allow scatter measurement at two angles this rudimentary approach cannot be used to separate populations of bioparticles of similar shape, size, or structure. The custom-built system used in the presented work collects axial light-loss and scatter signals at five carefully chosen angles. Experimental results obtained from colony scanner, as well from the extended cytometry instrument, were used to train the pattern-recognition algorithm. The results demonstrate that information provided by scatter alone may be sufficient to recognize various bioparticles with 90-99% success rate, both in flow and in imaging systems.