Fault‐tolerant spectroscopic data evaluation based on extended principal component regression correcting for spectral drifts and uncalibrated spectral features

Abstract

Chemometric algorithms are commonly applied in spectroscopy for deriving concentrations of chemicals from spectra. In order to ensure correct concentration results, the background spectrum must be stable and all absorbers present in unknown samples must have been contained in the calibration set. Recently, standard principal component regression (PCR) has been improved in two ways, correcting either for spectral drifts or for uncalibrated spectral features. In this study a combination of both correction approaches named secured pseudo PCR (spPCR) is presented which provides substantially improved concentration results even if the spectra are simultaneously affected by drifts and uncalibrated features. Additionally, spPCR extracts estimates of the spectrum of the uncalibrated absorbers and the drift spectrum. Hence systematic disturbances can be analyzed and a facile tool for allocating their origin is provided.

Interrelationships between generalized Tikhonov regularization, generalized net analyte signal, and generalized least squares for desensitizing a multivariate calibration to interferences

“…The interferent matrix L can include any artifact that the model needs to be desensitized to as in the orthogonal NAS processing examples previously noted (Andrew and Fearn, ; Westerhaus, ; Zhu et al ; Sun, ; Vogt and Mizaikoff, ; Ferré and Brown, ; Roger et al ; Shi and Anderson, ; Chen et al ; Soyemi et al ; Yang et al ; Yang et al ). It may be that improvements are possible in these references by using oblique adjustment via GNAS, GLS, or GTR.…”

“…Essential to both processes is a set of spectra spanning the interferent sample matrix, that is, spectra characterizing artifacts that the model needs to be desensitized to. Examples of orthogonal NAS pre‐processing applications include instrumentation differences (Andrew and Fearn, ; Westerhaus, ), replicate measurement variation such as in probe placement (Zhu et al ), baseline variations (Sun, ; Vogt and Mizaikoff, ), temperature (Ferré and Brown, ; Roger et al ), particle scattering (Shi and Anderson, ), optical path‐length correction (Chen et al ), and human subject differences in biomedical studies (Soyemi et al ; Yang et al ; Yang et al ).…”

Analytical Evaluation of Herbal Drugs

“…The other method is based on secured principal component regression (sPCR) that was originally developed for detecting and correcting uncalibrated spectral features newly emerging in spectra after the PCR calibration. It can detect and consider unexpected chromatographic features for quality valuation of herbal samples from the point of view of analyzing fingerprint residual Cheng & Chen, 2003;Collantes, 1997;Vogt, 2003;Welsh, 1996;Wold, 1977;Xie, 2001].…”

Support of Gulf of Mexico Hydrate Research Consortium: Activities to Support Establishment of a Sea Floor Monitoring Station Project

“…1 Significant strides in the development of a miniaturized deep-sea mid-infrared spectrometer ('sphere-IR') were completed in year 2 by establishing data evaluation strategies [2][3][4][5] , selection and testing of the fundamental instrument components including a Stirlingcooled MCT detector ideal for extended maintenance free deployment, and design considerations for an improved deep-sea sensor system from past experience of the PI's research group. [6][7][8][9][10][11] During year 3, computer assisted drawings (CAD) developed in year 2 for the primary instrument components facilitated the complete design and construction of the electrical compartment for 'SphereIR'.…”