Abstract
Background/Objectives: Two fibre optic probes were custom designed to perform Raman and near-infrared spectroscopic measurements. Our long-term objective is to develop a non-destructive tool able to collect data in hard-to-access locations for real-time analysis or diagnostic purposes. This study evaluated the quantitative performances of Probe A and Probe B using model pharmaceutical tablets. Methods: Measurements were performed using pharmaceutical tablets containing hydroxyl propylcellulose, titanium dioxide (anatase), lactose monohydrate, and indomethacin (γ form). Material content and thickness of bilayer samples (samples consisting of a top layer and a bottom layer of differing materials) were also assessed using Probe A to evaluate its capabilities to collect sub-surface information. Principal component analysis and partial least squares regression models were using individual and fused data to evaluate the performances of the different probe configurations. Results: Hydroxymethyl cellulose (RP2=0.98, RMSEP = 2.27% w/w) and lactose monohydrate (RP2=0.97, RMSEP = 2.96% w/w) content were most effectively estimated by near-infrared spectroscopy data collected using Probe A. Titanium dioxide (RP2=0.99, RMSEP = 0.21% w/w) content was most effectively estimated using a combination of 785 nm Raman spectroscopy and near-infrared spectroscopy using Probe B. Indomethacin (RP2=0.97, RMSEP = 1.01% w/w) was best estimated using a low-level fused dataset collected using 0 mm, 2.5 mm, and 5.0 mm lateral offsets of 785 nm spatially offset Raman spectroscopy using Probe A. Conclusions: The different probe configurations were able to reliably collect data and demonstrated robust quantitative performances. These results highlight the advantage of using multiple techniques for analysing different structures.