Progress in diagnostic pattern recognition (DPR)

Abstract

The identification and analysis of disease-specific signatures in mid-infrared spectra of serum forms the basis of a method called “Diagnostic Pattern Recognition (DPR)”. A conceivable usage of this method in clinical diagnostics requires that the method be applied in a convenient and robust manner. Thus, automation, room-temperature operation and reproducibility the prerequisite improvements toward routine application. We have investigated the performance two identical, semi-automated DPR systems. In contrast to previous measurements, which required MCT detectors, the use of a DLaTGS detector allowed the systems to be operated without the requirement of liquid nitrogen cooling. A series of measurements showed that automated pipetting improves the reproducibility significantly as compared to manual pipetting. For automated pipetting, the within-day variations are of minor importance. However, day-to-day variations may decrease the reproducibility in some spectral regions by more than a factor of two. Slight dependence of the reproducibility on the protein content of the serum samples has been observed.