Pretreatment Method Research of Near-Infrared Spectra in Blood Component Non-Invasive Measurement

Abstract

Blood component non-invasive measurement based on near-infrared (NIR) spectroscopy has become a favorite topic in the field of biomedicine. However, the various noises from instrument measurement and the varying background from absorption of other components (except target analyte) in blood are the main causes, which influenced the prediction accuracy of multivariable calibration. Thinking of backgrounds and noises are always found in high-scale approximation and low-scale detail coefficients. It is possible to identify them by wavelet transform (WT), which has multi-resolution trait and can break spectral signals into different frequency components retaining the same resolution as the original signal. Meanwhile, associating with a criterion of uninformative variable elimination (UVE), it is better to eliminate backgrounds and noises simultaneously and visually. Basic principle and application technology of this pretreatment method, wavelet transform with UVE criterion, were presented in this paper. Three experimental near-infrared spectra data sets, including aqueous solution with four components data sets, plasma data sets, body oral glucose tolerance test (OGTT) data sets, which, including glucose (the target analyte in this study), have all been used in this paper as examples to explain this pretreatment method. The effect of selected wavelength bands in the pretreatment process were discussed, and then the adaptability of different pretreatment method for the uncertainty complex NIR spectra model in blood component non-invasive measurements were also analyzed. This research indicates that the pretreatment methods of wavelet transform with UVE criterion can be used to eliminate varying backgrounds and noises for experimental NIR spectra data directly. Under the spectra area of 1100 to 1700 nm, utilizing this pretreatment method is helpful for us to get a more simple and higher precision multivariable calibration for blood glucose non-invasive measurement. Furthermore, by comparing with some other pretreatment methods, the results imply that the method applied in this study has more adaptability for the complex NIR spectra model. This study gives us another path for improving the blood component non-invasive measurement technique based on NIR spectroscopy.

Quantitative analysis of the hexamethylenetetramine concentration in a hexamethylenetetramine–acetic acid solution using near infrared spectroscopy: A comprehensive study on preprocessing methods and variable selection techniques

“…The reduction in the RMSEC and RMSEP values achieved through the application of the DWT algorithm highlights the efficacy of wavelet transformation in enhancing the spectral clarity and effectively removing unwanted noise components while retaining vital information. 29,30 This ultimately leads to improved model performance.…”

Fusion of various optimisation based feature smoothing methods for wearable and non‐invasive blood glucose estimation

“…Moreover, there are many substances in the blood vessel. The absorptions of the infrared light by these substances would cause the interferences to the obtained features [ 9 , 10 , 11 ]. Furthermore, there are the variations in the contact pressure and the contact position exerted on the sensor during the data acquisition.…”

A nonlinear model for calibration of blood glucose noninvasive measurement using near infrared spectroscopy

“…Currently, data preprocess is often utilized to eliminate noises or baseline. However, the research about elimination of matrix background interference in the field of blood glucose noninvasive measurement using NIR spectroscopy is very few [10]. In practice, the matrix background interference such as water is pivotal in NIR region for blood glucose noninvasive measurement.…”