Page 136 - Vibrational Spectroscopic Imaging for Biomedical Applications

112 Cha pte r F o u r

Two important factors to be kept in mind when using infrared
microspectroscopy for disease detection are that spectra with very
high signal-to-noise ratios (SNRs) are required and that those spectra
should be free from optical artifacts. Optical artifacts complicate spectral
interpretation and could prevent an accurate analysis and\or diagnosis.
Last, in any microspectroscopic analysis the sample itself becomes a
critical component in the optical system. Efforts to incorporate infrared
microanalysis into a protocol for disease detection have settled on the
38
use of low-E slides and TF analysis. These efforts have been the result
of combining two disparate disciplines, namely, infrared microspec-
troscopy and histology. While the preferred sample support for infrared
analysis is usually a hygroscopic alkali halide material, like sodium
chloride and potassium bromide, the histologist prefers glass substrates.
However, glass is not infrared transparent and the incorporation of alkali
halide materials into a histological preparation would be difficult, since
they are highly water soluble. Barium fluoride supports were initially
employed, but these are expensive and not in the microscope slide format
that the histologists prefer. A solution to these problems is low-E glass
slides which are transparent to visible light and reflecting for infrared light.
These slides are easily incorporated into any histological preparation and
allow the pathologist to view the sample using white-light microscopy
and the infrared microspectroscopist to study the sample in a TF analysis.
The only limitation is that the thickness of the tissue sample should be no
more than 6 μm. This thickness ensures that the features in an infrared
spectrum are not totally absorbing and that the spectra are photometri-
cally accurate from which quantitative data might be extracted.
In a typical TF analysis, light enters the sample from the objective at
an average incident angle of ~27°. The light transmits through the sam-
ple to the substrate, where it is reflected back through the sample and
collected by the objective. Based on Eq. (4.1), spectra from spatial domains
as small as 4λ (~24 μm for radiation possessing a wavelength of
6 micrometers) with SNRs of 1000/1 can be easily recorded. Further, the
average optical path length through the sample is 13.5 μm, based on the
sample thickness and incident angle given above. The analysis is straight
forward provided that the tissue sample is a continuous film possessing
low-contrast interfaces. Low-contrast interfaces are characterized as
those having optically similar materials present on either side of the inter-
face. Probably the most important parameter in this regard is the refrac-
tive index of both materials. However, the majority of tissue preparations
do not meet these criteria. Discontinuities within the sample (e.g., blood
vessels, vesicles, mineral inclusions) present interfaces with relatively
high-contrast edges. From an optical standpoint, a high-contrast edge can
promote scattering, diffraction, reflection, and dispersion. These effects
are further amplified due to the size and shape of the sample and the high
convergence of the impinging infrared radiation. Even worse, is the case
of a mineral inclusion which presents a high-contrast edge and a highly
scattering point defect. An additional artifact associated with this later