LF-NMR and multivariate data analysis: compression of data to classify hydrogel contact lenses

Hydrogel contact lenses swollen in viscoelastic artificial tears solution have been studied, measuring transversal relaxation times of water molecules using LF-NMR techniques. Data were processed by classical multiexponential fitting, by principal component analysis and by SLICING, a multi-way analysis method. The reason for using multivariate data analysis was not to obtain a better fitting, but rather more effective data description. The single-sample relaxation curves were projected in a space spanned by the loading curves, and in this space it was simpler to compare data. In particular, it contributed to the description of the variability of motion characteristics of the water molecule 'families' contained in the studied samples. Applying multivariate techniques, we were able to group lenses with different Equilibrium Water Content (EWC) and with the same water content but different compositions. Accordingly, we were able to point out that, if the lenses are swollen first in physiological solution and then in viscoelastic artificial tears solution, hydration characteristics remained unchanged in all the studied samples, except for 38% EWC lenses.