Majumder, S. K. ; Krishna, H. ; Gupta, P. K. ; Sidramesh, M. ; Chaturvedi, P. (2010) A comparative evaluation of Raman and fluorescence spectroscopy for optical diagnosis of oral neoplasia Proceedings of the SPIE, 8173 . 817306_1-817306_7. ISSN 0277-786X
Full text not available from this repository.
Official URL: http://link.aip.org/link/PSISDG/v8173/i1/p817306/s...
Related URL: http://dx.doi.org/10.1117/12.898904
Abstract
We report the results of a comparative evaluation of in vivo fluorescence and Raman spectroscopy for diagnosis of oral neoplasia. The study carried out at Tata Memorial Hospital, Mumbai, involved 26 healthy volunteers and 138 patients being screened for neoplasm of oral cavity. Spectral measurements were taken from multiple sites of abnormal as well as apparently uninvolved contra-lateral regions of the oral cavity in each patient. The different tissue sites investigated belonged to one of the four histopathology categories: 1) squamous cell carcinoma (SCC), 2) oral sub-mucous fibrosis (OSMF), 3) leukoplakia (LP) and 4) normal squamous tissue. A probability based multivariate statistical algorithm utilizing nonlinear Maximum Representation and Discrimination Feature for feature extraction and Sparse Multinomial Logistic Regression for classification was developed for direct multi-class classification in a leave-one-patient-out cross validation mode. The results reveal that the performance of Raman spectroscopy is considerably superior to that of fluorescence in stratifying the oral tissues into respective histopathologic categories. The best classification accuracy was observed to be 90%, 93%, 94%, and 89% for SCC, SMF, leukoplakia, and normal oral tissues, respectively, on the basis of leave-one-patient-out cross-validation, with an overall accuracy of 91%. However, when a binary classification was employed to distinguish spectra from all the SCC, SMF and leukoplakik tissue sites together from normal, fluorescence and Raman spectroscopy were seen to have almost comparable performances with Raman yielding marginally better classification accuracy of 98.5% as compared to 94% of fluorescence.
Item Type: | Article |
---|---|
Source: | Copyright of this article belongs to The International Society for Optical Engineering. |
ID Code: | 83739 |
Deposited On: | 22 Feb 2012 12:52 |
Last Modified: | 22 Feb 2012 12:52 |
Repository Staff Only: item control page