Please use this identifier to cite or link to this item:
Title: Classification of colonic tissues using raman spectroscopy and multivariate techniques
Authors: Huang, Z. 
Zheng, W. 
Widjaja, E.
Mo, J. 
Sheppard, C. 
Keywords: Adenocarcinoma
Colon cancer
Near-infrared Raman spectroscopy
Optical diagnosis
Support vector machines
Issue Date: 2006
Citation: Huang, Z., Zheng, W., Widjaja, E., Mo, J., Sheppard, C. (2006). Classification of colonic tissues using raman spectroscopy and multivariate techniques. Progress in Biomedical Optics and Imaging - Proceedings of SPIE 6093 : -. ScholarBank@NUS Repository.
Abstract: The purpose of this study was to explore the feasibility of using near-infrared (NIR) Raman spectroscopy and multivariate techniques for distinguishing cancer from normal and benign tissue in the colon. A total of 105 colonic specimens were used for Raman studies including 41 normal, 18 polyps, and 46 malignant tumors. The multivariate statistical techniques such as PCA-SVM were utilized to extract the significant Raman features and to develop effective diagnostic algorithms for tissue classification. The results showed that high-quality Raman spectra in the 800-1800 cm -1 range can be acquired from human colonic tissues in vitro, and Raman spectra differed significantly between normal, benign and malignant tumor tissue. PCA-SVM yielded a diagnostic sensitivity of 100%, 100%, and 97.7%, and specificity of 99.8%, 100%, and 100%, respectively, for differentiation between normal, polyp, and malignant tissue. Therefore, NIR Raman spectroscopy associated with multivariate techniques provides a significant potential for the noninvasive diagnosis of colonic cancers in vivo based on optical evaluation of biomolecules.
Source Title: Progress in Biomedical Optics and Imaging - Proceedings of SPIE
ISSN: 16057422
DOI: 10.1117/12.647384
Appears in Collections:Staff Publications

Show full item record
Files in This Item:
There are no files associated with this item.

Google ScholarTM



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.