Purchase this article with an account.
A. Nose, G. Tomita, S. T. Takagi, Y. Kita; Glaucoma Diagnosis in Japanese Eyes by Heidelberg Retina Tomography Using Japanese Database. Invest. Ophthalmol. Vis. Sci. 2008;49(13):3631.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Using advanced learning software (Relevance Vector Machine), Heidelberg Retina Tomograph (HRT; Heidelberg Engineering, Heidelberg, Germany) provides automatic, examiner independent discrimination between healthy and glaucomatous eyes (Glaucoma Probability Score, GPS). Recently, a database of Japanese eye (188 normative and 108 glaucomatous eyes) has been added and is able to be used for the GPS program. The purpose of this study was to assess the performance of the GPS and Moorfields Regression Analysis (MRA) based on the Japanese database and compared to that of the GPS and MRA based on the usual database.
The mean ± SD of the mean deviation of Humphrey visual field was - 6.04 ± 4.94dB (range: -26.16 to 0.52dB). Sensitivity and specificity for detecting glaucoma by GPS and MRA based on the Japanese database (J) was determined in 36 Japanese open-angle glaucoma eyes and 38 Japanese normative eyes. Only good quality images according to the automatic quality control of HRT were included in the study. The results were compared with those based on the Caucasian (C). A borderline was counted as "within normal limits". Analysis was also done by stratifying the eyes by refraction ( > -5.0 diopter or ≤ -5.0 diopter).
For glaucomatous eyes, sensitivity of GPS (72.2%) and MRA (77.8%) by the J database did not differ from that of GPS (86.1%) and MRA (88.9%) by the C database (Fisher’s exact tests, p > 0.1). Also, specificity of GPS (97.4%) and MRA (100%) by the J did not differ from GPS (92.1%) and MRA (84.2%) by the C. When stratified by the refractive error, sensitivity and specificity of GPS (J) and GPS (C) in eyes with refractive error ≤ -5.0 diopters (more myopic) was 50.0% / 100% and 68.8% / 100%, respectively (Fisher’s exact test, p >0.1).
At least in our series of Japanese eyes, diagnostic ability of GPS and MRA based on a Japanese database was not superior to that based on the Caucasian database, although no statistically significant difference was observed. At present, diagnosis of glaucoma by HRT may be more suitable using a mixed racial, large numbered database rather than using race specific one.
This PDF is available to Subscribers Only