Search

Recently proposed methods of bacteria identification in optical biosensors based on the phenomenon of light diffraction on macro-colonies offer over 98% classification accuracy. However, such high accuracy relies on the comparable and repeatable spatial intensity distribution of diffraction patterns. Therefore, it is essential to eliminate all non-species/strain-dependent factors affecting the diffraction patterns. In this study, the impact of the bacterial colony and illuminating beam misalignment on the variation of classification features extracted from diffraction patterns was examined. It was demonstrated that misalignment introduced by the scanning module significantly affected diffraction patterns and extracted classification features used for bacteria identification. Therefore, it is a crucial system-dependent factor limiting the identification accuracy. The acceptable misalignment level, when the accuracy and quality of the classification features are not affected, was determined as no greater than 50 µm. Obtained results led to development of image-processing algorithms for determination of the direction of misalignment and concurrent alignment of the bacterial colonies' diffraction patterns. The proposed algorithms enable the rigorous monitoring and controlling of the measurement's conditions in order to preserve the high accuracy of bacteria identification.

It is possible to identify bacteria species basing on their diffraction patterns followed by statistical analysis. The new approach exploits two steps: optimization of the recording conditions and introduction of new interpretable features for the identification. First, optimal diffraction registration plane, was determined. Next, results were verified by the analysis workflow based on ANOVA and Fisher divergence for feature selection, QDA and SVM models for classification and identification and CV with stratified sampling, sensitivity and specificity for performance assessment of the identification process. The proposed approach resulted in high sensitivity 0.9759 and specificity 0.9903 with very small identification error 1.34%.

It was demonstrated that statistical analysis of bacteria colonies Fresnel patterns recorded in the optical system with converging spherical wave illumination is suitable for highly effective bacteria species classification. The proposed method includes Fresnel patterns recording followed by image processing and the statistical analysis based on feature extraction, feature selection, classification and classification performance methods. Examination performed on various bacteria species (Salmonella enteritidis, Staphylococcus aureus, Staphylococcus intermedius, Escherichia coli, Proteus mirabilis, Pseudomonas aeruginosa and Citrobacter freundii) revealed that the proposed method achieved very high accuracy of over 98%.