Purchase this article with an account.
T. Mumcuoglu, E. Wu, G. Wollstein, H. Ishikawa, R.A. Bilonick, L. Kagemann, M.L. Gabriele, T. Capozzoli, R.J. Noecker, J.S. Schuman; Macular Retinal Segmentation and Composite Retinal Image for Glaucoma Detection . Invest. Ophthalmol. Vis. Sci. 2006;47(13):3356.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
We have previously shown that macular retinal segmentation analysis of optical coherence tomography (OCT) macular scans discriminates between healthy and glaucomatous eyes (IOVS 2005; 46:2012–7). The purpose of this study was to investigate whether applying the segmentation analysis on a composite of retinal scans further improves discrimination.
Twenty–one eyes (11 healthy and 10 glaucomatous subjects) had undilated Stratus OCT fast macular scans performed 3 consecutive times in a single visit. Composite images were created from the 3 macular scans, using software of our own design. Scans were aligned to the internal limiting membrane and the mean reflectivity value was calculated for each corresponding pixel. Retinal segmentation analysis was applied to the macular scans before and after creating the composite image. Four retinal layers were defined: macular nerve fiber layer (mNFL), inner retinal complex (IRC; retinal ganglion cell layer+inner plexiform and nuclear layers), outer plexiform layer (OPL), and outer retinal complex (ORC; outer nuclear layer+photoreceptor layer). All parameters of macular segmentation before and after the composite processing, and avarege macular retinal thickness from the standard analysis of Stratus OCT were recorded. Area under the receiver operator characteristics (AROC) were calculated for discrimination between healthy and glaucomatous eyes.
The mean visual field mean deviation (MD) was –0.31 ± 0.93 dB for the healthy group and –2.63 ± 2.37 dB for the glaucoma group. AROC for the mean macular retinal thickness was 0.85. Applying the retinal segmentation improved the discrimination capabilities with the highest AROC for mNFL (0.95) followed by mNFL+IRC (0.88). Further improvement was noted when the segmentation analysis was applied on the composite image with AROC=1 for mNFL followed by mNFL+IRC (0.92).
Macular segmentation of a composite retinal image improves the ability to distinguish between healthy and glaucomatous eyes when compared to mean macular retinal thickness.
This PDF is available to Subscribers Only