Advanced Spectroscopy Technique for Biomedicine - Biomedical Optical Imaging Technologies: Design and Applications - page 43

gastric mucosa at 1,452, 1,522, or 1;660 cm 1 . The intensity ratio of 1;156 cm 1 to
the maximum peak and 1;587-1;156 cm 1 could be used to differentiate malignant
from normal tissue with an accuracy of 100%. Kalyan Kumar et al. [ 76 ] studied
normal and malignant gastric tissues using a commercial Raman microspectroscopy
at 785-nm excitation. A total of 111 spectra (37 normal and 74 malignant) from
10 normal and 17 malignant stomach mucosae were obtained. PCA and LDA
analysis showed a 93% of sensitivity and 84% of specificity. Kawabata et al. [ 77 ]
studied the gastric cancer tissues with a 1,064-nm excitation. They found that based
on the 1;644 cm 1 peak alone, an accuracy of 70% could be realized.
In vivo gastric cancer diagnosis using endoscopic Raman spectroscopy has been
implemented by Huang et al. [ 78 , 79 ] for the first time. The probe is shown in
Fig. 1.26 . It is integrated with trimodal wide-field imaging for guiding the in vivo
gastric tissue Raman measurements. Similar to the lung Raman probe of [ 39 , 40 ],
the endoscopic probe for gastric cancer diagnosis consists of 33 ultralow-OH fibers
(NA D 0:22, 200-m core diameter). The center fiber, which is coated with a
narrow 785 band-pass filter, is for excitation laser delivery, and the surrounding
32 fibers, which are coated with 785-nm long-pass filter, are used for signal
collection. Both the excitation and signal collection are filtered using a second
stage in-line filter module. High S/N Raman spectra could be obtained within 1 s.
Preliminary in vivo Raman results showed that the ratio at 875-1; 450 cm 1 could
be used to differentiate dysplasia from normal gastric tissue with a sensitivity and
specificity of 100% [ 79 ]. Very recently, the authors studied gastric dysplasia using
narrow-band image-guided Raman spectroscopy [ 80 ]. It demonstrated the potential
of the endoscopic Raman spectroscopy for in vivo gastric malignancy diagnosis.

1.4.6 Breast Cancer Diagnosis

Breast cancer is one of the most common cancers for women. It is estimated
that there are about 216,000 new cases each year and 40,000 died of the disease.
The common method is clinical examination or mammography followed by an
invasive biopsy. Optical spectroscopic techniques such as fluorescence, reflectance,
and Raman spectroscopy have already been used for breast cancer diagnosis [ 81 ].
Trends correlating with diseases have been identified from fluorescence spectra of
ex vivo breast tissue samples [ 82 , 83 ]. Because there are only a couple of endogenous
fluorophores that can be detectable and the spectra shape is relatively broad, the
sensitivity and specificity are relatively low. Raman spectroscopy is found to be
promising in breast cancer diagnosis. Early studies were based on small sample
sets and ex vivo measurements [ 84 - 89 ]. A typical Raman spectrum of normal
human breast tissue is shown in Fig. 1.27 [ 85 ]. The Raman spectrum is heavily
dependent on the excitation wavelength. Spectra of formalin-fixed human tissue
revealed Raman peaks for lipids and carotenoids [ 90 ]. The lipid features were
better defined under 782- and 830-nm laser excitation, while the carotenoid features
were strongest under 488- and 515-nm excitation due to resonance enhancement.