Speaker
Description
Laser-induced breakdown spectroscopy (LIBS) is a real-time spectrochemical technique that involves performing time-dependent optical emission spectroscopy on high-temperature laser-induced microplasmas. The use of a focused laser beam allows one to make sensitive assays of the elemental composition of specimens while requiring only micrograms of analyte mass. Numerous medical and biomedical applications of LIBS have been studied and proposed, including the rapid detection of bacterial infection.
Our group has demonstrated the ability to detect and classify bacterial cells in arbitrary fluid specimens using a centrifugation device to deposit cells on nitrocellulose filters. Bacterial concentrations of 11,000 CFU per laser shot were detectable using 8 mJ pulses from a 1064 nm, 9 ns Nd:YAG laser with a spot size 75 micron in diameter. A partial least squares discriminant analysis on LIBS spectra from specimens of blood and urine spiked with known bacterial pathogens possessed a 98.9% sensitivity and 100% specificity for detection in urine and a 96.3% sensitivity and 98.6% specificity for detection in blood.
An artificial neural network analysis with principle component analysis pre-processing of the LIBS spectrum was used to discriminate three species of bacteria. Use of an 80:20 split cross-validation resulted in an average sensitivity and specificity of 97.2% and 98.6%, respectively, for the discrimination of bacteria in urine. External validation performed on 16 filters gave an average sensitivity of 77.5%. Applying PCA-ANN using an 80:20 split cross-validation for the discrimination of bacteria in blood resulted in 100% sensitivity and specificity. External validation of 19 filters of bacteria in blood yielded an average sensitivity of 82.3%. These results indicate the potential usefulness of LIBS in the clinical setting.
Work supported by the Natural Sciences and Engineering Research Council (NSERC) of Canada.
| Keyword-1 | optical spectroscopy |
|---|---|
| Keyword-2 | pathogen diagnostic |
| Keyword-3 | artificial neural network |
