Stalked protozoa identification by image analysis and multivariable statistical techniques.

Protozoa are considered good indicators of the treatment quality in activated sludge systems as they are sensitive to physical, chemical and operational processes. Therefore, it is possible to correlate the predominance of certain species or groups and several operational parameters of the plant. This work presents a semiautomatic image analysis procedure for the recognition of the stalked protozoa species most frequently found in wastewater treatment plants by determining the geometrical, morphological and signature data and subsequent processing by discriminant analysis and neural network techniques. Geometrical descriptors were found to be responsible for the best identification ability and the identification of the crucial Opercularia and Vorticella microstoma microorganisms provided some degree of confidence to establish their presence in wastewater treatment plants.