| |
107. DOG-Kongress Home
DOG-Kongressinformation
DOG-Kongress Bildergalerie
Grußworte
Organisation, Termine
Ablauf des Kongresses
Preise und Forschungsförderungen
Höhepunkte
Wissenschaftliches Programm
Feierliche Eröffnung
Schwerpunkte
Wissenschaftliches Programm
- Do, 24.09.09
- Fr, 25.09.09
- Sa, 26.09.09
- So, 27.09.09
- Posterausstellung
Symposien
Kurse
Firmenveranstaltungen
Satellitenprogramm
Hinweise, Informationen
Rahmenprogramm
Sponsoren, Industrie
Presseservice
Programm downloaden / drucken [PDF, 11 MB]
Vorprogramm downloaden / drucken [PDF, 3 MB]
DOG-Homepage
|
|
Abstract
SO.16.02
Influence of retinal blood vessels on diagnostic accuracy of RNFL thickness measurements using spectral-domain OCT to detect glaucoma
Christian Mardin, Folkert Horn, Delia Baleanu, Anselm Jünemann, Robert Lämmer, Friedrich Kruse, Ralf-Peter Tornow
Universitäts-Augenklinik Erlangen
Objective
To measure the potential influence of retinal blood vessels (rbv) on the diagnostic accuracy of RNFL thickness measurements to detect glaucoma.
Methods
Circular B-Scans (diameter 3.4 mm, 768 A-Scans) around the optic nerve head (ONH) were acquired in 99 healthy subjects and 48 Glaucoma patients using spectral domain OCT (Spectralis HRA & OCT, Heidelberg Engineering) and RNFL thickness was measured. B-Scan images were exported and the position of rbv was calculated by locating the rbvs’ shadows in the layer of the pigment epithelium using a customized software (see fig.). RNFL thickness was set to zero at positions were rbv were found. Two data sets were used for further calculations: one with rbv included in the RNFL thickness results and one without rbv. The thickness values were averaged to 6 sectors (with 0° corresponding temporal, horizontal): temporal (315 – 34 deg), temporal-superior TS (34 – 79 deg), superior S (79 – 124 deg), nasal N (124 – 225 deg), inferior I (225 – 270 deg), temporal-inferior TI (270 – 315 deg). Diagnostic accuracy was investigated using ROC-curves for mean and sectoral RNFL thickness.
Results
Mean thickness in normal subjects was (mean+SD) (99+9) µm and (88+8) µm with and without vessels, respectively. This resulted in a mean thickness of the rbv layer of 11 µm corresponding to 11 % of the total RNFL. The fraction of the rbv contribution to RNFL thickness in the 6 sectors was highest superiorly and temporally: T: (2±3) %, TS: (15±7) %, S: (15±9) %, N: (9±4) %, I: (14±7) %, TI: (15±6) %. There was no difference in the area under the ROC curves for mean RNFL thickness (0.982 vs 0.983) with and without rbv and in the 6 sectors.
Conclusions
The exclusion of rbv from RNFL thickness measurements using spectral domain OCT does not affect the diagnostic accuracy of mean RNFL thickness measurements in this data set. Sector data show differences depending on sector position. |
|