Uwe Damm

Spectral Variable Selection for Partial Least Squares Calibration Applied to Authentication and Quantification of Extra Virgin Olive Oils Using Fourier Transform Raman Spectroscopy

The limits of quantitative multivariate assays for the analysis of extra virgin olive oil samples from various Greek sites adulterated by sunflower oil have been evaluated based on their Fourier transform (FT) Raman spectra. Different strategies for wavelength selection were tested for calculating optimal partial least squares (PLS) models. Compared to the full spectrum methods previously applied, the optimum standard error of prediction (SEP) for the sunflower oil concentrations in spiked olive oil samples could be significantly reduced. One efficient approach (PMMS, pair-wise minima and maxima selection) used a special variable selection strategy based on a pair-wise consideration of significant respective minima and maxima of PLS regression vectors, calculated for broad spectral intervals and a low number of PLS factors. PMMS provided robust calibration models with a small number of variables. On the other hand, the Tabu search strategy recently published (search process guided by restrictions leading to Tabu list) achieved lower SEP values but at the cost of extensive computing time when searching for a global minimum and less robust calibration models. Robustness was tested by using packages of ten and twenty randomly selected samples within cross-validation for calculating independent prediction values. The best SEP values for a one years harvest with a total number of 66 Cretian samples were obtained by such spectral variable optimized PLS calibration models using leave-20-out cross-validation (values between 0.5 and 0.7% by weight). For the more complex population of olive oil samples from all over Greece (total number of 92 samples), results were between 0.7 and 0.9% by weight with a cross-validation sample package size of 20. Notably, the calibration method with Tabu variable selection has been shown to be a valid chemometric approach by which a single model can be applied with a low SEP of 1.4% for olive oil samples across three different harvest years.

Bedside monitoring of subcutaneous interstitial glucose in healthy individuals using microdialysis and infrared spectrometry

An IR-spectroscopy-based bedside device, coupled to a subcutaneously implanted microdialysis probe, is developed for quasicontinuous glucose monitoring with intermittent readouts at 10-min intervals, avoiding any sensor recalibration under long-term operation. The simultaneous estimation of the microdialysis recovery rate is possible using an acetate containing perfusate and determining its losses across the dialysis membrane. Measurements are carried out on four subjects, with experiments lasting either 8 or 28 h, respectively. Using the spectral interval data either from 1180 to 950 or 1560 to 1000 cm(-1), standard errors of prediction (SEPs) between 0.13 and 0.28 mM are achieved using multivariate calibration with partial least-squares (PLS) or classical least-squares (CLS) calibration models, respectively. The transfer of a PLS calibration model using the spectral and reference concentration data of the dialysates from the three 8-h-long experiments to a 28-h monitoring episode with another healthy subject is tested. Including microdialysis recovery for the determination of the interstitial glucose concentrations, an SEP of 0.24 mM is obtained versus whole blood glucose values. The option to determine other metabolites such as urea or lactate offers the possibility to develop a calibration- and reagent-free point-of-care analyzer.

Glucose Quantification in Dried-down Nanoliter Samples Using Mid-Infrared Attenuated Total Reflection Spectroscopy

The aim of this study was to determine the feasibility of minimally invasive glucose concentration measurement of a body fluid within the physiologically important range below 100 nL with a number of samples such as interstitial fluid, plasma, or whole blood using mid-infrared spectroscopy, but starting with preliminary measurements on samples of simple aqueous glucose solutions. The Fourier transform infrared spectrometer was equipped with a Golden Gate® single reflection diamond attenuated total reflection (ATR) accessory and a room-temperature pyroelectric detector. As the necessary detection limits can be achieved only for dried samples within the spectrometric conditions realized by a commercial instrument, the work focused on the optimization of such ATR measurements. We achieved quantification of samples with volumes as low as 7 nL between 10 and 600 mg/dL. The standard error of prediction (SEP) for the concentration range 10–100 mg/dL is 3.2 mg/dL with full interval data between 1180 and 940 cm−1. The performance of the prediction is given by a coefficient of variation of prediction (CVpred) of 6.2%. When all samples within the whole concentration range are included, the SEP increases to 20.2 mg/dL, and hence the CVpred to 10.6% due to a nonlinear signal dependence on glucose concentration. A detection limit for glucose of 0.7 ng with a signal-to-noise ratio of 10 was obtained.

Infrared Transmission Spectrometry for the Determination of Urea in Microliter Sample Volumes of Blood Plasma Dialysates

Application of mid-infrared spectroscopy for the determination of urea in blood plasma dialysates of microliter sample volumes using a transmission microcell was investigated. Infrared spectra of the dialysates of plasma samples collected from 75 different patients using CMA 60 microdialysis catheters were evaluated with multivariate partial least squares regression. Using the absorbance spectral data from 1520–1420 cm−1 and 1220–1120 cm−1, a minimum standard error of prediction (SEP) of 0.88 mg/dL (0.14 mM) was achieved with spectral variable selection. Our findings suggest the feasibility of developing a mid-infrared sensor in combination with micro-fluidics for on-line monitoring of urea in patients undergoing dialysis treatment.

Transmission infrared spectroscopy of whole blood – complications for quantitative analysis from leucocyte adhesion during continuous monitoring

Infrared spectroscopy has been applied to analyse glucose and cellular components in whole blood with the aim of developing an online clinical diagnostic and monitoring modality. Leucocyte adsorption onto the CaF(2) windows was observed over a period of several hours under continuous blood flow using a transmission cell of 30 mum path length. This build-up of cellular material on the windows is responsible for diminishing the sample path length under the flow conditions chosen. The adsorption dynamics have been characterised and their impact on glucose monitoring is reported. For short-term monitoring (<2 hours) a standard error of prediction of 11 mg/dL with human citrated blood samples from three different subjects was achieved. Furthermore, the leucocyte build-up was also reported for porcine EDTA blood monitoring. Consequences and testing opportunities with regard to the first stages in the immune cell reaction to the exposure of body-foreign materials to anticoagulated whole blood are discussed.

Microdialysis based monitoring of subcutaneous interstitial and venous blood glucose in Type 1 diabetic subjects by mid-infrared spectrometry for intensive insulin therapy

Implementing strict glycemic control can reduce the risk of serious complications in both diabetic and critically ill patients. For this purpose, many different blood glucose monitoring techniques and insulin infusion strategies have been tested towards the realization of an artificial pancreas under closed loop control. In contrast to competing subcutaneously implanted electrochemical biosensors, microdialysis based systems for sampling body fluids from either the interstitial adipose tissue compartment or from venous blood have been developed, which allow an ex-vivo glucose monitoring by mid-infrared spectrometry. For the first option, a commercially available, subcutaneously inserted CMA 60 microdialysis catheter has been used routinely. The vascular body interface includes a double-lumen venous catheter in combination with whole blood dilution using a heparin solution. The diluted whole blood is transported to a flow-through dialysis cell, where the harvesting of analytes across the microdialysis membrane takes place at high recovery rates. The dialysate is continuously transported to the IR-sensor. Ex-vivo measurements were conducted on type-1 diabetic subjects lasting up to 28 hours. Experiments have shown excellent agreement between the sensor readout and the reference blood glucose concentration values. The simultaneous assessment of dialysis recovery rates renders a reliable quantification of whole blood concentrations of glucose and metabolites (urea, lactate etc) after taking blood dilution into account. Our results from transmission spectrometry indicate, that the developed bed-side device enables reliable long-term glucose monitoring with reagent- and calibration-free operation.

Bedside monitoring of subcutaneous interstitial glucose in type 1 diabetic subjects using microdialysis and infrared spectrometry with optimal correlation to blood glucose concentrations

Infrared spectroscopy has been successfully employed in multi-component assays for the study of various biomedical samples. Two areas have found particular interest, i.e. in-vitro analysis in the clinical laboratory and point-of-care applications. With regard to the latter field, in-vivo blood glucose monitoring is an important topic for improving glycemic control in critically ill patients with non-adequate blood glucose regulation, similar to the situation faced for diabetic patients. For such application, a continuously operated mid-infrared spectroscopic system in combination with a subcutaneously implanted microdialysis probe and coupled by micro-fluidics has been developed. Using the dialysis process, the interstitial fluid matrix can be significantly simplified, since high molecular mass compounds such as proteins are separated. However, the micro-dialysis recovery rate is variable over time, so that a simultaneous determination of this parameter was implemented using the losses of an acetate marker from the perfusate across the dialysis membrane. Clinical measurements were carried out on type 1 diabetic subjects, with experiments lasting up to 28 hours. The concentrations of glucose, acetate and other components in the dialysates from interstitial body fluids were investigated. Two different multivariate calibration strategies, i.e. partial least squares (PLS) and classical least squares (CLS) regressions were applied. The results led to excellent correlation of the subcutaneous interstitial concentrations with those of laboratory blood glucose readings. Clarke-Error-Grid evaluations were employed for assessing the clinical applicability of the method.

Reliable long-term continuous blood glucose monitoring for patients in critical care using microdialysis and infrared spectrometry

For clinical research, in-vivo blood glucose monitoring is an ongoing important topic to improve glycemic control in patients with non-adequate blood glucose regulation. Critically ill patients received much interest, since the intensive insulin therapy treatment, as established for diabetics, reduces mortality significantly. Despite the existence of commercially available, mainly amperometric biosensors, continued interest is in infrared spectroscopic techniques for reagent-free glucose monitoring. For stable long-term operation, avoiding also sensor recalibration, a bed-side device coupled to a micro-dialysis probe was developed for quasi-continuous glucose monitoring. Multivariate calibration is required for glucose concentration prediction due to the complex composition of dialysates from interstitial body fluid. Measurements were carried out with different test persons, each experiment lasting for more than 8 hours. Owing to low dialysis recovery rates, glucose concentrations in the dialysates were between 0.83 and 4.44 mM. Standard errors of prediction (SEP) obtained with Partial Least Squares (PLS) calibration and different cross-validation strategies were mainly between 0.13 and 0.18 mM based on either full interval data or specially selected spectral variables.