Advanced Spectroscopy Technique for Biomedicine - Biomedical Optical Imaging Technologies: Design and Applications - page 27

its corresponding SUM. Setting DEV D 0 is equivalent to Lieber's method except
an additional peak removal procedure, but applying our rule provides a means for
taking into account the noise effect and avoiding artificial peaks that may arise
from noise and from both ends of the spectra. In order to minimize the distortion
of the polynomial fitting by major Raman signals, the major peaks are identified
and are removed from the following rounds of fitting. Peak removal is limited
to the first few iterations to prevent unnecessary excessive data rejection. The
iterative polynomial fitting procedure is terminated when further iterations cannot
significantly improve the fitting, determined by j .DEV i DEV i 1 /=DEV i j <5%.
As with many iterative computation methods, the percentage can be empirically
adjusted by the user according to the problem involved and computation time
allowed. The final polynomial fit is regarded as the fluorescence background. The
final Raman spectra are derived from the raw spectra by subtracting the final
polynomial fit function. A copy of the algorithm for noncommercial use can be
downloaded from http://www.bccrc.ca/dept/ic/cancer-imaging/haishan-zeng-phd .

1.3.2 Real-Time Raman Spectroscopic System for Endoscopic

Lung Cancer Detection

Lung cancers have the highest mortality rate among all cancers and are second only
to skin cancers in incidence. It is estimated that nearly 170,000 Americans died of
lung and bronchial cancers each year. The overall 5-year survival rate of patients
with lung cancers is around 15%, much lower than patients with other types of
cancers. An effective way to reduce lung cancer mortality rates is early detection
followed by surgery and other therapies. In the past decade, autofluorescence
spectroscopy and imaging were developed for in vivo early detection of lung cancers
[ 35 - 37 ]. However, the spectral features of tissue autofluorescence are broad and
show less specific differences between normal and pathologic sites. In an earlier
in vitro study [ 38 ], we found that the Raman spectra of lung cancer were greatly
different from those of the normal lung tissue. This prompts us to design a real-time
Raman spectroscopic system for improving in vivo lung cancer detection [ 39 , 40 ].
A few technical difficulties must be overcome in designing spectroscopic system
for endoscopic applications: (1) The size of the Raman probe is limited by the
size of the instrument channel of an endoscope. Although it is desirable to have
a large probe to collect as much signal as possible, it must be small enough to pass
through the instrument channel. (2) The measurement time is limited by the clinical
procedure. The measurement time allowed for an in vivo application is less than a
couple of seconds. (3) Raman signal of lung tissue is very weak. Real-time in vivo
Raman imaging of macroscopic tissue area is technologically infeasible. Therefore,
Raman spectroscopy itself cannot be used as an efficient method for locating the
lesion. We have to rely on autofluorescence or white light imaging to locate the sites
of concern and then apply the probe for Raman spectra measurement. The real-
time in vivo endoscopic Raman system consisted of an excitation light source,