L E Pablo1, A Ferreras, A B Pajarín, P Fogagnolo. 1. Department of Ophthalmology, Miguel Servet University Hospital, Instituto Aragonés de Ciencias de la Salud, and University of Zaragoza, Zaragoza, Spain.
Abstract
BACKGROUND/AIMS: To calculate and validate a linear discriminant function (LDF) for optical coherence tomography (OCT) to improve the diagnostic ability of isolated optic nerve head (ONH) parameters to discriminate between healthy individuals and glaucoma patients. METHODS: Two independent samples (teaching and validating sets) were prospectively selected. The teaching set (54 normal eyes and 73 glaucoma patients) was used to calculate the LDF. The validating set (70 healthy individuals and 67 glaucoma patients) was used to test the performance of the LDF in an independent population. Receiver operating characteristic (ROC) curves were plotted and compared with the ONH parameters measured using OCT. RESULTS: The optimized function was: LDF=8.204+(2.59xdisc area)-(9.25xhorizontal rim width). The largest areas under the ROC curve were 0.923 and 0.898 for our LDF and the vertical rim area in the validating population respectively. There were no significant differences between the areas. At 95% fixed specificity, the LDF (71.64%) and the vertical rim area (65.67%) yielded the highest sensitivity values. CONCLUSIONS: All ONH parameters obtained with OCT, except disc area, had a good ability to differentiate between healthy and glaucoma individuals. As judged by the area under the ROC curve, the LDF performed better than any single parameter, although statistics did not prove it better than vertical cup/disc ratio or vertical rim area with the number of subjects we studied.
BACKGROUND/AIMS: To calculate and validate a linear discriminant function (LDF) for optical coherence tomography (OCT) to improve the diagnostic ability of isolated optic nerve head (ONH) parameters to discriminate between healthy individuals and glaucomapatients. METHODS: Two independent samples (teaching and validating sets) were prospectively selected. The teaching set (54 normal eyes and 73 glaucomapatients) was used to calculate the LDF. The validating set (70 healthy individuals and 67 glaucomapatients) was used to test the performance of the LDF in an independent population. Receiver operating characteristic (ROC) curves were plotted and compared with the ONH parameters measured using OCT. RESULTS: The optimized function was: LDF=8.204+(2.59xdisc area)-(9.25xhorizontal rim width). The largest areas under the ROC curve were 0.923 and 0.898 for our LDF and the vertical rim area in the validating population respectively. There were no significant differences between the areas. At 95% fixed specificity, the LDF (71.64%) and the vertical rim area (65.67%) yielded the highest sensitivity values. CONCLUSIONS: All ONH parameters obtained with OCT, except disc area, had a good ability to differentiate between healthy and glaucoma individuals. As judged by the area under the ROC curve, the LDF performed better than any single parameter, although statistics did not prove it better than vertical cup/disc ratio or vertical rim area with the number of subjects we studied.
Authors: Ashish Sharma; Jonathan D Oakley; Joyce C Schiffman; Donald L Budenz; Douglas R Anderson Journal: Ophthalmology Date: 2011-03-12 Impact factor: 12.079
Authors: Beatriz Abadia; Antonio Ferreras; Pilar Calvo; Mirian Ara; Blanca Ferrandez; Sofia Otin; Paolo Frezzotti; Luis E Pablo; Michele Figus Journal: Biomed Res Int Date: 2014-06-16 Impact factor: 3.411
Authors: Pilar Calvo; Antonio Ferreras; Beatriz Abadia; Mirian Ara; Michele Figus; Luis E Pablo; Paolo Frezzotti Journal: Biomed Res Int Date: 2014-07-06 Impact factor: 3.411
Authors: Jean-Claude Mwanza; Gary Lee; Donald L Budenz; Joshua L Warren; Michael Wall; Paul H Artes; Thomas M Callan; John G Flanagan Journal: Transl Vis Sci Technol Date: 2018-04-03 Impact factor: 3.283