Objective. To determine the potential utility of polarized light used during colposcopic examinations.
Materials and Methods. Matched, polarized, and unpolarized colposcopic images and diagnostic annotations from 31 subjects receiving excisional treatment of cervical neoplasia were compared. Sensitivity, specificity, and mean Euclidean distances between the centroids of the gaussian ellipsoids for the different epithelial types were calculated for unpolarized and polarized images.
Results. The sensitivities of polarized colposcopic annotations for discriminating cervical intraepithelial neoplasia (CIN) 2 or higher were greater for all 3 acetowhite categories when compared with unpolarized annotations (58% [44/76] vs 45% [34/76], 68% [50/74] vs 59% [45/76], and 68% [49/72] vs 66% [50/76], respectively). The average percent differences in Euclidean distances between the epithelial types for unpolarized and polarized cervical images were as follows: CIN 2/3 versus CIN 1 = 33% (10/30, p =.03), CIN 2/3 versus columnar epithelium = 22% (p =.004), CIN 2/3 versus immature metaplasia = 29% (14/47, p =.11), and CIN 1 versus immature metaplasia = 27% (4.4/16, p =.16).
Conclusions. Because of its ability to interrogate at a deeper plane and eliminate obscuring glare, polarized light colposcopy may enhance the evaluation and detection of cervical neoplasias.
Use of polarized light may improve colposcopic examinations.
1Gynecologic Cancer Prevention Center, Departments of Family Medicine and Obstetrics and Gynecology, Medical College of Georgia, Augusta, GA; 2STI Medical Systems, Honolulu, HI; 3The University of Michigan, Ann Arbor, MI; and 4Instituto Nacional de Enfermedades Neoplasicas, Lima, Peru
Correspondence to: Daron G. Ferris, MD, Gynecologic Cancer Prevention Center, 1423 Harper St, Medical College of Georgia, Augusta, GA 30912. E-mail: firstname.lastname@example.org
Grants awarded to Instituto Nacional de Enfermedades Neoplasicas, Lima, Peru, and Medical College of Georgia Research Institute by STI helped support this work.