8 May 2000 Recognition of disease-specific patterns in FT-IR spectra of human sera
Author Affiliations +
Proceedings Volume 3918, Biomedical Spectroscopy: Vibrational Spectroscopy and Other Novel Techniques; (2000) https://doi.org/10.1117/12.384942
Event: BiOS 2000 The International Symposium on Biomedical Optics, 2000, San Jose, CA, United States
Vibrational spectra in the mid-IR region show significant and reproducible correlation with the disease state of the blood donor. When focusing our 'disease pattern recognition (DPR)' approach onto the example of diabetes mellitus we can clearly separate samples obtained from healthy volunteers from those samples which organized from diabetes patients. Furthermore, we are able to differentiate between samples of type-1 diabetics and type-2 diabetics. For disease pattern recognition we use linear and/or regularized discriminant analysis. In a binary, supervised classification of an pair of the three disease states: healthy, diabetes type-1 and diabetes type-2, we consistently achieve sensitivities and specificities >= 80 percent. By setting stricter bounds on the range of acceptable probabilities of belonging to a certain class, we obtain even higher values for the sensitivity and the specificity on the expense of the fraction of 'crisply' classified samples. Since we are able to simultaneously quantify the concentrations of biochemical serum components like glucose, cholesterol and triglycerides from the identical set of spectra with regression coefficients > 90 percent, our approach allows for a direct cross-link between the molecule-based and the disease-based interpretation of the spectra.
© (2000) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Wolfgang H. Petrich, Wolfgang H. Petrich, Brion Dolenko, Brion Dolenko, Johanna Frueh, Johanna Frueh, Helmut Greger, Helmut Greger, Stephan Jacob, Stephan Jacob, Franz Keller, Franz Keller, Alexander Nikulin, Alexander Nikulin, Matthias Otto, Matthias Otto, Ortrud Quarder, Ortrud Quarder, Raymond L. Somorjai, Raymond L. Somorjai, Arnulf Staib, Arnulf Staib, Gerhard H. Werner, Gerhard H. Werner, Hans Wielinger, Hans Wielinger, } "Recognition of disease-specific patterns in FT-IR spectra of human sera", Proc. SPIE 3918, Biomedical Spectroscopy: Vibrational Spectroscopy and Other Novel Techniques, (8 May 2000); doi: 10.1117/12.384942; https://doi.org/10.1117/12.384942

Back to Top