Improved Principal Component Analysis (IPCA): A Novel Method for Quantitative Calibration Transfer between Different...

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Chemometrics

Background:

• Model transfer in near-infrared (NIR) spectroscopy is crucial due to the labor-intensive nature of model establishment.
• Existing methods primarily focus on model transfer between identical instrument types.
• A significant challenge remains in transferring models between different NIR spectrometers with varying spectral characteristics.

Purpose of the Study:

• To develop a novel methodology for calibration transfer between different types of NIR spectrometers.
• To expand the applicability of NIR spectroscopic models across diverse instruments.
• To address the limitations of current methods in handling spectral variations from different spectrometers.

Main Methods:

• Proposed a new methodology based on improved principal component analysis (IPCA) for calibration transfer.
• Evaluated the IPCA method using three datasets, including public pharmaceutical tablets, corn data, and FT-NIR/MicroNIR spectrometer data.
• Compared IPCA performance against the piecewise direct standardization (PDS) method.

Main Results:

• IPCA demonstrated comparable results to PDS for spectra from the same instrument type.
• IPCA achieved successful calibration transfer between benchtop (FT-NIR) and micro (MicroNIR) spectrometers, even with wavelength selection.
• The proposed IPCA method improved prediction ability without degrading model performance compared to original micro spectra.

Conclusions:

• The IPCA method effectively enables calibration transfer between different types of NIR instruments.
• This approach overcomes limitations related to spectral differences (wavenumber, absorbance) between spectrometers.
• IPCA offers a versatile supporting technology for the practical application of NIR spectroscopy across a wide range of instruments.