Identificacion Espectroscopica en la Region Media y Cercana del Infrarrojo para su Aplicacion en Mediciones No Invasivas de Glucosa

摘要

Invasiveness of actual glucose measuring methods for diabetic patients has impulsed studies of non-invasive techniques, within which infrared (IR) spectroscopy has become one of the most accepted approaches. However, there is a disagreement between studies that identify measuring characteristics in near infrared (NIR) or mid infrared (MIR) regions. In this work a quantitative analysis in both regions is presented in order to identify important peaks and parts that could contain information related to glucose concentration. 12 solutions of glucose diluted in ultrapure water were prepared, in order to cover the clinical range of glucose concentration in human blood (40mg/dL - 400mg/dL). Transmittance spec-trums in NIR and MIR regions were collected, and different wavelengths and regions that could be related with glucose concentration were identified. 22 calibration models were built with algorithms like partial least squares regression (PLS) and principal component regression (PCR). The models were compared by using the standard error of calibration (SEC), the standard error of prediction (SEP) and a contrast test of significance. Important information was identified in the MIR region between 990 cm"1 and 1130 cm"1 with a peak in 1042 cm"1, and in the NIR region between 7700 cm"1 and 4000 cm"1. These results will be used in further analysis with more complex matrixes as plasma, whole blood and some tissue simulators, in order to identify interferences in identified regions.