Page 301 - Vibrational Spectroscopic Imaging for Biomedical Applications

Raman Imaging for Biomedical Applications in Clinics 275

For point and line Raman mapping approaches, the duration of the
experiment depends critically on the laser excitation power density and
the number of points to be scanned. The experimental comparison dem-
onstrates that line mapping represents the fastest method for acquiring
spectral information at a reasonable spatial resolution, typically 1 μm,
while yielding reconstructed Raman images of good quality.
In contrast to the two mapping approaches, the time needed for
global imaging depends primarily on the number of spectral chan-
nels at which an entire image is recorded. Global Raman imaging
employing LCTF technology is the method of choice for obtaining
sub-micrometer, essentially diffraction-limited, spatially resolved
high-quality images from flat samples. To characterize a sample’s
chemical heterogeneity, only a relatively few global Raman images
need to be recorded at well-defined wavenumber positions, which
are known either a priori or from a spectral analysis of data obtained
in point or line scanning determinations. 28
Further development of LCTF will lead to better transmission so
smaller accumulation times for increased S/N ratio that will make
that method more used in wider applications. However, this tech-
nique is not optimal for full spectral imaging as most generated
Raman photons are rejected by the filter and only a small part is
imaged on the detector.
There are other methods that try to enhance existing techniques.
Hadamard transform imaging also employs global illumination, as in
direct imaging. In this method, a series of images encoded with
Hadamard transform mask (binary spatial light modulation) are
recorded, and from these images high-definition and high-spatial-
resolution Raman images can be reconstructed. 27
Another approach is the fiber-bundle image compression method
that uses a two-dimensional array of fibers to collect a Raman spectra
from a grid of sample points. The fibers can then be arranged in a
linear array and imaged on the CCD so that all three dimensions (two
spatial and one spectral) are being measured simultaneously. 29
Although this is an attractive method to obtain immediate Raman
images, the imaging on fiber bundles results in a rather large signal
loss compared to normal line imaging.
Therefore, if one wants to obtain high-resolution Raman spectra
with a high resolution, line imaging seems the most efficient method.
28
As mentioned, there are three parameters to consider, the signal acqui-
sition time, the exposure time and the signal-to-noise ratio of the
spectra. For low-intensity signals, the total signal exposure time in
combination with the readout speed determines the amount of noise
in the spectra. Therefore, longer integration times per pixel are used,
in combination with a slow readout speed, resulting in long collection
times per image. For high-intensity signals, as is the case in Hadamard
transform imaging, the integration time can be much shorter and the
readout speed higher. Although this seems attractive, it should be noted