Quantitative aement of atypical birefringence image uing canning laer polarimetry with variable corneal compenation

Purpoe To define the clinical characteritic of atypical birefringence image and to decribe a quantitative method for their identification. Deign Propective, comparative, clinical obervational tudy. Method Normal and glaucomatou eye underwent complete examination, tandard automated perimetry, canning laer polarimetry with variable corneal compenation (GDx-VCC), and optical coherence tomography (OCT) of the macula, peripapillary retinal nerve fiber layer (RNFL), and optic dik. Eye were claified into two group: normal birefringence pattern (NBP) and atypical birefringence pattern (ABP). Clinical, functional, and tructural characteritic were aeed eparately. A multiple logitic regreion model wa ued to predict eye with ABP on the bai of a quantitative can core generated by a upport vector machine (VM) with GDx-VCC. Reult ixty-five eye of 65 patient were enrolled. ABP image were oberved in 5 of 20 (25%) normal eye and 23 of 45 (51%) glaucomatou eye. Compared with eye with NBP, glaucomatou eye with ABP demontrated ignificantly lower VM core (P < .0001, < 0.0001, 0.008, 0.03, and 0.03, repectively) and greater temporal, mean, inferior, and naal RNFL thickne uing GDx-VCC; and a weaker correlation with OCT generated RNFL thickne (R2 = .75 v .27). ABP image were ignificantly correlated with older age (R2 = .16, P = .001). The VM core wa the only ignificant (P < .0001) predictor of ABP image and provided high dicriminating power between eye with NBP and ABP (area under the receiver operator characteritic curve = 0.98). Concluion ABP image exit in a ubet of normal and glaucomatou eye, are aociated with older patient age, and produce an artifactual increae in RNFL thickne uing GDx-VCC. The VM core i highly predictive of ABP image.