Share Email Print
cover

Proceedings Paper

White-light oblique-incidence diffuse reflectance spectroscopy for classification of in-vivo pigmented skin lesions
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

A study of in-vivo classification of pigmented skin lesions using oblique-incidence diffuse reflectance spectroscopy is presented. Spatio-spectral data in the wavelength range from 455 to 765 nm are collected from 111 pigmented lesions including 10 histopathologically diagnosed as melanoma. The first 67 lesions are used for training the classifiers, and 44 lesions are used for testing. The first classifier separates (1) malignant melanoma and severe dysplastic nevi from (2) moderate and mild dysplastic nevi, common nevi, actinic and seborrheic keratoses. The second classifier next distinguishes between (a) moderate and mild dysplastic nevi, common nevi from (b) actinic and seborrheic keratoses. The third classifier further separates (I) moderate and mild dysplastic nevi from (II) common nevi. The first classifier performs with 100% sensitivity and 91% specificity with overall classification rates of 93% and 95 % for the training and testing sets, respectively. The second classifier has classification rates of 95% and 97 % for the training and testing sets, respectively, whereas the third classifier has classification rates of 98% and 94 % for the training and testing sets, respectively.

Paper Details

Date Published: 21 February 2007
PDF: 6 pages
Proc. SPIE 6435, Optical Interactions with Tissue and Cells XVIII, 64350L (21 February 2007); doi: 10.1117/12.699633
Show Author Affiliations
Alejandro Garcia-Uribe, Texas A&M Univ. (United States)
Elizabeth B. Smith, Texas A&M Univ. (United States)
Madeleine Duvic, The Univ. of Texas, M.D. Anderson Cancer Ctr. (United States)
Lihong V. Wang, Washington Univ. in St.Louis (United States)


Published in SPIE Proceedings Vol. 6435:
Optical Interactions with Tissue and Cells XVIII
Steven L. Jacques; William P. Roach, Editor(s)

© SPIE. Terms of Use
Back to Top