Abstract
Abstract: :
Purpose: To detect angle-closure glaucoma using biometric data obtained by Orbscan and an artificial neural network.Methods: 180 normal eyes (including eyes with ametropia) and 20 phakic eyes of 20 patients with a history of acute angle-closure glaucoma in the fellow eye were included. Biometric measurements obtained by Orbscan. (i.e., iridocorneal angle and anterior chamber depth) and A scan echography (i.e., lens thickness and axial length) were recorded. We compared biometric measurements in both groups and we created an artificial neural network using Orbscan biometric measurements. Results: Both groups were significantly different (p<0.001) for the iridocorneal angle (43.8° versus 30.1°), the anterior chamber depth (3.04 mm versus 2.26 mm), the lens thickness (3.96 mm versus 5.03 mm), and for the axial length (23.1 mm versus 22.2 mm). An artificial neural network using age, sex, subjective spherical equivalent, iridocorneal angle, and anterior chamber depth allowed a correct classification in 95% of the eyes. The sensibility and the specificity of this model were respectively 90% and 95.5%. Conclusion: Iridocorneal angle and anterior chamber depth measured by Orbscan can be analyzed using in an artificial neural network to detect angle-closure glaucoma.
Keywords: 354 clinical (human) or epidemiologic studies: prevalence/incidence • 318 anterior segment • 354 clinical (human) or epidemiologic studies: prevalence/incidence