Advanced Spectroscopy Technique for Biomedicine - Biomedical Optical Imaging Technologies: Design and Applications - page 39

OH stretching motions of water. In vivo lung Raman spectra of different pathologies
showed clear difference, particularly in the HF range. The malignant lesions (SCC
and CIS) could be easily differentiated from the benign lesion (moderate dysplasia)
and normal tissue. These findings indicated that there was a real possibility that
in vivo lung tissue could be characterized at the molecular level using Raman
spectroscopy. Currently, we are conducting a large-scale clinical study to test
the utility of the real-time Raman spectroscopic system for in vivo lung cancer
detection.

1.4.3 Colon Cancer Diagnosis

Colon cancer is the third most common form of cancer and the second leading
cause of cancer-related deaths in the Western world. The golden standard for colon
cancer diagnosis is white light endoscopy followed by invasive biopsy. However,
the ability of white light endoscopy to visually detect dysplasia in precancerous
conditions or reliably differentiate dysplastic from nondysplastic colon polyps
is very limited. Optical spectroscopy techniques including diffuse reflectance,
fluorescence, and Raman spectroscopy have been evaluated for diagnosis [ 63 ]. Shim
et al. reported the measurement of colon cancer tissues in vivo and ex vivo using
a fiber-based Raman probe [ 47 , 64 ]. For the ex vivo study, a total of 33 polyps
were analyzed from 8 patients. Fifty-four spectra were obtained from the 33 polyps
including 20 hyperplastic and 34 adenomatous. For the ex vivo measurement, each
spectrum was collected with an integration time of 30 s. For in vivo measurement,
19 spectra were collected from 9 polyps in 3 patients including 9 hyperplastic
and 10 adenomatous. Each spectrum was collected with an integration time of
5 s. A PCA and LDA analysis found that adenomatous polyps could be identified
with 91% sensitivity, 95% specificity, and 93% accuracy for ex vivo studies and
100% sensitivity, 89% specificity, and 95% accuracy for in vivo studies. Recently,
Widjaja et al. studied the capability of support vector machines for classification
of the near-infrared Raman spectra of in vitro colonic tissues [ 65 ]. A total of
105 colonic tissue specimens from 59 patients including 41 normal, 18 hyperplastic,
and 46 adenocarcinomas were studied. A total of 817 spectra including 324 normal,
184 polyps, and 309 adenocarcinomatous were studied using the conventional
support vector machine (SVM). An overall accuracy of over 99% was achieved.
The Raman spectra of normal colonic tissue were also studied by FT-Raman
spectroscopy [ 66 ], microconfocal Raman spectroscopy, and FTIR spectroscopy
[ 67 ]. Typical Raman spectra of the normal, precancerous, and cancerous colonic
tissue are shown in Fig. 1.22 . The primary Raman peaks for normal, benign, and
malignant tumors are around 875, 1,002, 1,090, 1,267, 1,320, 1,445, 1,605, 1,655,
and 1;740 cm 1 . Significant difference between the tissue types including peak
intensity, peak position, and spectral shapes could also be easily observed, indicating
the changes of biochemical composition and structures of the tissue associated with
the disease state.