PURPOSE: To evaluate the accuracy of a new objective method based on Placido disk-derived data for the detection of eyes at risk of ectasia. METHODS: One hundred nineteen eyes of 176 patients were included and separated into two groups, normal and forme fruste keratoconus (FFKC), using automated corneal classification software. Normal eyes (n = 114) were classified as negative for keratoconus and keratoconus suspect and had undergone LASIK with unremarkable follow-up for 4 years. The FFKC group was composed of 62 topographically normal eyes of patients with keratoconus in the fellow eye. Anterior topographic parameters, obtained from specular topography using Placido-based indices and corneal wavefront Zernike coefficients, were compared between groups. Receiver operating characteristic (ROC) curves were used to identify cut-off points in discriminating between keratoconic and normal eyes. Validation was performed on an external group of eyes. RESULTS: A discriminant function was built combining four corneal wavefront variables and four Placido variables. The area under the receiver operating characteristic was 0.970 with this eight-variable model. The validation of this function had 63% sensitivity for detecting FFKC and 100% sensitivity for detecting keratoconus, with a specificity of 82%. CONCLUSIONS: Indices generated from corneal wavefront and Placido measurements can assist in identifying early or mild forms of keratoconus undetected by a Placido-based neural network program. [J Refract Surg. 2016;32(8):510-516.]. Copyright 2016, SLACK Incorporated.
PURPOSE: To evaluate the accuracy of a new objective method based on Placido disk-derived data for the detection of eyes at risk of ectasia. METHODS: One hundred nineteen eyes of 176 patients were included and separated into two groups, normal and forme fruste keratoconus (FFKC), using automated corneal classification software. Normal eyes (n = 114) were classified as negative for keratoconus and keratoconus suspect and had undergone LASIK with unremarkable follow-up for 4 years. The FFKC group was composed of 62 topographically normal eyes of patients with keratoconus in the fellow eye. Anterior topographic parameters, obtained from specular topography using Placido-based indices and corneal wavefront Zernike coefficients, were compared between groups. Receiver operating characteristic (ROC) curves were used to identify cut-off points in discriminating between keratoconic and normal eyes. Validation was performed on an external group of eyes. RESULTS: A discriminant function was built combining four corneal wavefront variables and four Placido variables. The area under the receiver operating characteristic was 0.970 with this eight-variable model. The validation of this function had 63% sensitivity for detecting FFKC and 100% sensitivity for detecting keratoconus, with a specificity of 82%. CONCLUSIONS: Indices generated from corneal wavefront and Placido measurements can assist in identifying early or mild forms of keratoconus undetected by a Placido-based neural network program. [J Refract Surg. 2016;32(8):510-516.]. Copyright 2016, SLACK Incorporated.
Authors: Oren Golan; Andre L Piccinini; Eric S Hwang; Ildamaris Montes De Oca Gonzalez; Mark Krauthammer; Sumitra S Khandelwal; David Smadja; J Bradley Randleman Journal: Am J Ophthalmol Date: 2019-02-02 Impact factor: 5.258