Submitted to: Biosensors World Congress
Publication Type: Abstract Only
Publication Acceptance Date: 5/17/2002
Publication Date: N/A
Citation: N/A Interpretive Summary:
Technical Abstract: The low selectivity of microbial biosensors often limits the field of their practical use. Application of pattern recognition methods for processing data of sensor arrays can effectively solve this problem. The aim of this study was to demonstrate the possibility of determining the composition of a multicomponent mixture using a measuring system based solely on an array of non-selective microbial biosensors. Glucose (10g/1), xylose (30g/1) and ethanol (1 g/1) were chosen as model substrates. Measurements were carried out using an amperometric flow- injection system, consisting of three microbial biosensors (based on Hansenula polymorpha, Escherichia coli and Gluconobacter oxydans strains). Cluster analysis was used for data processing. Calibration of the system revealed eight clusters of signals corresponding to all possible combina- tions of the three substrates. Recognition of a sample composition was based on the Euclidean distance from the sample response to the center of the appropriate cluster. Cluster analysis allowed successful identifica- tion of 37 of 39 ternary mixture samples (analyzed to determine the eight combinations). Thus, a pattern recognition method used traditionally for the analysis of chemical sensor data was successfully extended to biosensor analysis. Further development of this method will make it possible to determine the concentrations of each substrate and to increase the number of detectable components.