Validation of a natural urinary stone data based infrared library searching system with artificial survey samples.

In this study we evaluated a Fourier transform IR spectrometer (Bio-Rad, USA) equipped with a search system for the analysis of urinary stones. We constructed a database of the stone library with the help of results found with X-ray diffraction analysis. In total, we included 223 stone results (213 natural and 10 spurious stones) consisting of single and composite nature. Regarding the latter we used many comparable and many diverging combinations. Applying 60 artificial referee samples that were used in the urinary stone surveys as organized by the German Society for Clinical Chemistry, we found the instrument hit quality index alone, as a measure of best spectral match not entirely sufficient in relation to acceptable performance. This was also caused by the absence of some rare components. Of those 60 survey samples, 17 did not meet the guidelines of the survey organization. These guidelines are: qualitatively correct and quantitatively within tolerance limits +/-20%. Though a number of these samples seemed irrelevant in a clinical setting, they nevertheless represent a challenge for better performance. In the second round, therefore, we included new entries and human expertise as well, which resulted in an upgrading of the score. We only missed, finally, 4 combinations, mainly related to the purine molecule, i.e. uric acid, sodium urate and ammonium urate. In conclusion, based on extension of the library, we consider the search system as acceptable. Despite that, human interpretation proved to be necessary.