Control and data analysis tool for a novel FT-NIR spectrometer

Near-infrared (NIR) spectroscopy is being applied to the solution of problems in many areas of biotechnology and pharmaceutical industry. The need for modern medical diagnostics to develop small portable instruments that enable fast and effective monitoring of the biological properties of the human body is apparent. We have developed a portable and robust spectrometer that consists of a two-beam interferometer operating in the NIR wavelength range for real-time measurements. An active control system has been designed for the automatic alignment of the instrument´s optical components. One of the mirrors is mounted upon three piezoelectric transducers (PZT) arranged so that they can both tilt and retard the mirror. Laser sources and corresponding photo-detectors are also incorporated such that the control system can use the PZT actuators to produce a series of mirror movements relative to the fixed mirror. During the scanning process the control system keeps the mirrors accurately aligned in real time, as the interferometer is scanning and also while it is idling between scans. This is known as dynamic alignment. The reliability and repeatability performance of the alignment system were tested with spectral calibration experiments. For the analysis, advanced signal processing has been applied to NIR data collected from the new device to estimate the instrument´s ability to produce precise and informative diagnostic information. Our target application concerns blood and tissue status in a form that can be interpreted directly by the user, without special knowledge of spectral analysis. More specifically, dimensionality reduction techniques and artificial neural networks were first applied to measure different concentrations of urea and creatinine samples in vitro. The results are encouraging, with overall mean squared prediction errors on the order of 10^-5, and in vivo trials will follow to further develop the device and the techniques