Biomedical Engineering Reference
In-Depth Information
in this end that is connected to a white light source. Before each measurement, this
fiber is imaged onto the skin (by turning on the white light source) to help locate the
measurement position. Once the position is decided, the white light source is turned
off before Raman measurement.
The end of the fiber bundle that is coupled to the spectrograph has a proprietary
design that substantially improves the signal-to-noise ratio (S/N) of the Raman
system and is discussed in detail in Sect. 1.3.1.6 .
1.3.1.5
Raman Signal Detection
The detection unit is equipped with an NIR-optimized back-illuminated deep-
depletion CCD array (LN/CCD-1024EHRB, Princeton Instruments, Trenton, NJ)
and a transmissive imaging spectrograph (HoloSpec-f /2.2-NIR, Kaiser, Ann Arbor,
MI) with a holographic grating (HSG-785-LF, Kaiser, Ann Arbor, MI). The CCD
has a 16-bit dynamic range and is liquid nitrogen-cooled to 120 ı C. The f -number
of the spectrograph (f /2.2) matches the numerical aperture (NA D 0:22)ofthe
fiber, so the throughput is much better than that of a traditional f /4 Czerny-Turner
spectrographs [ 32 ]. The spectral resolution of the system with 100-mfibersis
8 cm 1 .
1.3.1.6
Aberration Correction and Hardware Binning
As discussed in Sect. 1.2.5.3 , conventional spectrograph is equipped with a straight
slit, and the image of the straight slit is parabolic through a plane grating. For
spectrographs with short focal lengths, this distortion is significant that can affect the
performance of spectral detection. The horizontal displacement of a spectral line of
the real-time Raman system is shown graphically in Fig. 1.12 , with the displacement
rounded to pixels (dashed lines). The maximum horizontal displacement is 5 pixels
(135 m). The solid line is a linear-regression-fitted parabolic curve described by
x D 1:1904 10 5 y 2
1:9455 10 4 y 0:98613;
(1.9)
where x is the horizontal displacement at a vertical position, y. Manufacturer
(Kaiser) of the HoloSpec spectrograph recommended a combination of hardware
and software binning in which the neighboring pixels along the dashed vertical line
are hardware binned, then shifted to the appropriate number of pixels and summing
them up with software. As shown in Fig. 1.12 , there are 11 such hardware-binning
groups. Complete software binning can also be used to correct the aberration, in
which the whole image is acquired first and then all the pixels along the curve are
added up together using software. However, the improvement of S/N ratio using
software binning or combined hardware-software binning is limited because the
binning only improves the S/N ratio by as much as the square root of the number
of pixels binned together. For weak Raman signals that are readout noise limited,
hardware binning is preferred because it improves the S/N ratio linearly with the
Search WWH ::




Custom Search