16 February 2012 Multivariate analysis methods for spectroscopic blood analysis
Author Affiliations +
Abstract
Blood tests are an essential tool in clinical medicine with the ability diagnosis or monitor various diseases and conditions; however, the complexities of these measurements currently restrict them to a laboratory setting. P&P Optica has developed and currently produces patented high performance spectrometers and is developing a spectrometer-based system for rapid reagent-free blood analysis. An important aspect of this analysis is the need to extract the analyte specific information from the measured signal such that the analyte concentrations can be determined. To this end, advanced chemometric methods are currently being investigated and have been tested using simulated spectra. A blood plasma model was used to generate Raman, near infrared, and optical rotatory dispersion spectra with glucose as the target analyte. The potential of combined chemometric techniques, where multiple spectroscopy modalities are used in a single regression model to improve the prediction ability was investigated using unfold partial least squares and multiblock partial least squares. Results show improvement in the predictions of glucose levels using the combined methods and demonstrate potential for multiblock chemometrics in spectroscopic blood analysis.
© (2012) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Michael F. G. Wood, Michael F. G. Wood, Arash Rohani, Arash Rohani, Rashid Ghazalah, Rashid Ghazalah, I. Alex Vitkin, I. Alex Vitkin, Romuald Pawluczyk, Romuald Pawluczyk, } "Multivariate analysis methods for spectroscopic blood analysis", Proc. SPIE 8219, Biomedical Vibrational Spectroscopy V: Advances in Research and Industry, 821909 (16 February 2012); doi: 10.1117/12.908324; https://doi.org/10.1117/12.908324
PROCEEDINGS
9 PAGES


SHARE
Back to Top