Blood Glucose Determination by Fourier Transform near Infrared Spectroscopy

Abstract

: Diabetes is a metabolic disorder that is caused by unregulated blood glucose and therefore requires regular and intensive monitoring. Currently, blood sugar monitoring is mostly done invasively by withdrawing blood through a needle or piercing of the fingertips. This method can cause trauma and an infection. However, there is the potential for using a non-invasive measurement of blood glucose levels with near-infrared spectroscopy (NIRS) combined with partial least-square regression. As a pathway to actualize it, the spectrum of whole blood was measured with different glucose levels. A total of 72 NIR spectrum from 8 whole blood samples with different types of glucose levels were measured. A principal component analysis (PCA) and partial least square regression (PLSR) were applied to the spectral data matrix. The results showed that PCA is successfully classified as spectral data based on the glucose content and PLSR model within the clinically accurate region of the Clarke error grid. These results indicate that NIRS has an immense potential to be applied in measuring blood glucose non-invasively.