Abstract
Abstract: :
Purpose: To study the investigator-dependent diagnostic variability of optic nerve head measurements with confocal laser scanning tomography (HRT I). Methods: A HRT I (Heidelberg Engineering GmbH, Heidelberg) database with 50 HRT images from 25 adults was independently analysed by 7 masked, experienced ophthalmologists from 6 eye-centres. Images were obtained from healthy volunteers and glaucoma patients. Only refractive error, keratometry and age of the patients were known to the observers. Each examiner was asked to diagnose glaucoma from the HRT I images of the optic nerve heads (ONHs). Inter-, and intra-observer differences were calculated as mean±s.d. of disc area, cup area and other parameters for each investigator using software HRT 2.01. Statistical analysis was performed by means of SPSS (SPSS inc., Chicago). Results: Intra-observer means for disc area [mm2] (n=50) varied from user 1 to 7 (random order) 1.83±0.49 / 2.10±0.46 / 2.14±0.36 / 1.92±0.45 / 2.05±0.49 / 2.06±0.41 / 2.21±0.40 with a mean inter-observer variation of 8.8% (range: 2.3-50.6%). Intra-observer means for cup area were 0.45±0.35 / 0.46±0.37 / 0.43±0.35 / 0.45±0.35 / 0.45±0.35 / 0.44±0.37 / 0.44±0.37 with a mean inter-observer variation of 16.9% (range: 0.7-63.6%). Glaucoma was correctly diagnosed by 2 observers only. 24 healthy ONHs were suspected for glaucoma. Conclusion: The variation of intra-observer mean values of HRT I measurements was higher for disc area than for cup area. The mean inter-observer variation for cup area was greater. The observer dependent diagnostic variability of HRT I measurements may lead to disagreeing results. Our findings confirm that the artificial colour coded HRT I image analysis is not suitable for detecting glaucomatous ONHs when used as single diagnostic method.
Keywords: 430 imaging/image analysis: clinical • 432 imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • 498 optic disc