ARTIFICIAL NEURAL-NETWORK PREDICTIONS OF URINARY CALCULUS COMPOSITIONS ANALYZED WITH INFRARED-SPECTROSCOPY

Infrared (IR) spectroscopy is used to analyze urinary calculus (renal stone) constituents. However, interpretation of IR spectra for quantif ying urinary calculus constituents in mixtures is difficult, requiring expert knowledge by trained technicians. In our laboratory IR spectra of unknown calculi are compared with reference spectra in a computeri zed library search of 235 reference spectra from various mixtures of c onstituents in different proportions, followed by visual interpretatio n of band intensities for more precise semiquantitative determination of the composition. To minimize the need for this last step, we tested artificial neural network models for detecting the most frequently oc curring compositions of urinary catculi. Using constrained mixture des igns, we prepared various samples containing ammonium hydrogen urate, brushite, carbonate apatite, cystine, struvite, uric acid, weddellite, and whewellite for use as a training set. We assayed known artificial mixtures as well as selected patients' samples from which the semiqua ntitative compositions were determined by computerized library search followed by visual interpretation. Neural network analysis was more ac curate than the library search and required less expert knowledge beca use careful visual inspection of the band intensities could be omitted . We conclude that neural networks are promising tools for routine qua ntification of urinary calculus compositions and for other related typ es of analyses in the clinical laboratory.