Light scattering during infrared spectroscopic measurements of biomedical samples

摘要

Infrared (IR) spectroscopy has shown potential to quickly and non-destructively measure the chemical signatures of biomedical samples such as single biological cells, and tissue from biopsy. The size of a single cell (diameter ~10-50 µm) are of a similar magnitude to the mid-IR wavelengths of light (~1-10 µm) giving rise to Mie-type scattering. The result of this scattering is that chemical information is significantly distorted in the IR spectrum.Distortions in biomedical IR spectra are often observed as a broad oscillating baseline on which the absorbance spectrum is superimposed. A spectral feature commonly observed is the sharp decrease in intensity at approximately 1700 cm-1, next to the Amide I band (~1655 cm-1), which pre-2009 was called the 'dispersion artefact'. The first contributing factor towards the 'dispersion artefact' investigated was the reflection signal arising from the air to sample interface entering the collection optics during transflection experiments. This was theoretically modelled, and then experimentally verified. It was shown that IR mapping could be done using reflection mode, yielding information from the optically dense nucleus which previously caused extinction of light in transmission mode.The most important contribution to the spectral distortions was due to resonant Mie scattering (RMieS) which occurs when the scattering particle is strongly absorbing such as biomedical samples. RMieS was shown to explain both the baselines in IR spectra, and the 'dispersion artefact' and was validated using a model system of poly(methyl methacrylate) (PMMA) of varying sizes from 5 to 15 µm. Theoretical simulations and experimental data had an excellent match thus proving the theory proposed. With an understanding of the physics/mathematics of the spectral distortions, a correction algorithm was written, the RMieS extended multiplicative signal correction (RMieS-EMSC). This algorithm modelled the measured spectrum as superposition of a first guess (the reference spectrum) which was of a similar biochemical composition to the pure absorbance spectrum of the sample, and a scattering curve. The scattering curve was estimated as the linear combination of a database of a large number of scattering curves covering a range of feasible physical parameters. Simulated and measured data verified that the RMieS-EMSC increased IR spectral quality.