Chemical sensing technology is being developed for applications in biomedical research, clinical chemistry, neuroscience, bioprocess monitoring, biotechnology, and environmental sciences.  Spectroscopic and optical methods are used to measure selected chemical analytes in complex matrices.  Examples include noninvasive blood glucose sensing in people with diabetes, on-line systemic urea monitoring during hemodialysis treatments, real-time monitoring of cellular nutrients and metabolites in functioning bioreactors, and remote, continuous dissolved oxygen measurements in biological systems.

      Near infrared spectroscopy is being developed to measure blood glucose levels in human subjects both painlessly and noninvasively.  The concept is to pass a harmless band of near infrared light through the human body and then extract the glucose concentration information from the transmitted spectral radiation.  Success of this approach is based on the unique spectral absorption features of various biological species.  The following figure illustrates the differences in the near infrared spectrum for a selected group of molecules of clinical significance.  Although unique, these spectral features are heavily overlapping and weak in magnitude.   These properties require high signal-to-noise ratios and complex data analysis algorithms to extract the desired analytical information in a selective and robust manner.  Digital filtering techniques and multivariate calibration methods, such as partial least-squares regression, are generally used to extract glucose information accurately from noninvasive near infrared spectra collected from complex biological matrices. 

      We have developed an in vitro model of the human body to help identify and characterize experimental parameters that are critical for noninvasive blood glucose measurements.  Investigated parameters include scattering, spectrometer signal-to-noise ratios, optical path length and spectral correlations within a data set. 

Recently, we have succeeded in measuring glucose noninvasively from human subjects by an analysis of spectra collected across tongues.  Although measurement errors are too large for clinical purposes, these experimental results demonstrate the possibility of noninvasive blood glucose measurements.  Current research is focused on improving both accuracy and precision of these noninvasive measurements by enhancing the signal-to-noise of the required instrumentation.  In addition to blood glucose, we are presently exploring the application of near infrared spectroscopy for measuring: 1) systemic urea during hemodialysis treatments, 2) glucose, lactate, ammonia, and glutamine during fermentation of mammalian and insect cells, and 3) dissolved protein during crystallization.

      Many sensor applications demand continuous operation over extended periods without user intervention.  The measurement of dissolved oxygen in growth media during bioreactor operation on-board the Space Shuttle is one such application.  Sensor recalibration is not possible after the shuttle is launched, which means measurement accuracy requires stable calibrations.  Stable chemical sensors are being developed for remote sensing by using radioluminescent (RL) light sources.  Our RL sources are composed of tritium gas coupled with an inorganic phosphor.  Beta emission from the tritium excites a phosphor and the excited phosphor subsequently relaxes to produces a photon.   RL-optical sensors for dissolved oxygen are being developed on the basis of dynamic fluorescence quenching of immobilized ruthenium complexes.  Low noise and high stability of RL sources provide low detection limits and long term stability. 

The above figure presents actual oxygen measurements made while growing mammalian brain cells during the Columbia Space Shuttle flight (STS-93).  Pre- and post-launch data is presented for oxygen levels before and after the reaction.   Temperature and humidity levels are also indicated.   A photograph of the flight unit is provided below.   This unit consists of eight sensors for monitoring before and after four independent reactors.   Four of the eight input/output ports are visible on the photographed end plate.   The remaining ports are located on the opposite end plat, which is identical.  

Kromoscopy is a new measurement code for analytical science.   In this method, white light passes through the sample and the transmitted light is divided into four separate detector channels.  The response function of each channel is defined by the source, detector, and bandpass function of a filter that is position immediately before the detector.  Each chemical species displays a unique Kromoscopic response when represented as a vector in the multidimensional space defined by the four detector signals.  The above figure shows a three-dimensional vector plot which illustrates the difference in response for a selected group of biomolecules.   The unique direction for glucose relative to the other primary components of whole blood is particularly noteworthy.  In addition to selectivity, this instrument configuration provides tremendously high signal-to-noise ratios, which are required for noninvasive analytical measurements.  We are presently characterizing the selectivity of Kromoscopy by defining conditions under which glucose can be measured at clinically relevant concentrations in the presence of common, endogenous interferenes, such as urea, triglycerides, hemoglobin, cholesterol, and protein.

Examples include noninvasive blood glucose sensing in people with diabetes, on-line systemic urea monitoring during hemodialysis treatments, real-time monitoring of cellular nutrients and metabolites in functioning bioreactors, and remote, continuous dissolved oxygen measurements in biological systems.