gms | German Medical Science

26. Jahrestagung der Deutschen Retinologischen Gesellschaft

Deutsche Gesellschaft für Retinologie

27.09.2013, Hamburg

Posterior Pole Choroidal Vasculature Assessed By Automated Choroidal Vessel Segmentation In Standard Clinical Spectral-Domain Optical Coherence Tomography In Patients With Diabetic Macular Edema

Meeting Abstract

  • Bianca Gerendas - Medizinische Universtität Wien, Austria
  • S.M. Waldstein - Medizinische Universtität Wien, Austria
  • B. Hajnajeeb - Medizinische Universtität Wien, Austria
  • L. Zhang - Department of Electrical and Computer Engineering, University of Iowa, Iowa City, USA
  • H. Bogunovic - Department of Electrical and Computer Engineering, University of Iowa, Iowa City, USA
  • M. Abramoff - Department of Ophthalmology & Visual Sciences, University of Iowa Hospital & Clinics, Iowa City, USA
  • C. Simader - Medizinische Universtität Wien, Austria
  • M. Sonka - Department of Electrical and Computer Engineering, University of Iowa, Iowa City, USA; Department of Ophthalmology & Visual Sciences, University of Iowa Hospital & Clinics, Iowa City, USA
  • U. Schmidt-Erfurth - Medizinische Universtität Wien, Austria

Retinologische Gesellschaft. 26. Jahrestagung der Retinologischen Gesellschaft. Hamburg, 27.-27.09.2013. Düsseldorf: German Medical Science GMS Publishing House; 2013. Doc13rg26

doi: 10.3205/13rg26, urn:nbn:de:0183-13rg268

Veröffentlicht: 20. August 2013

© 2013 Gerendas et al.
Dieser Artikel ist ein Open Access-Artikel und steht unter den Creative Commons Lizenzbedingungen (http://creativecommons.org/licenses/by-nc-nd/3.0/deed.de). Er darf vervielfältigt, verbreitet und öffentlich zugänglich gemacht werden, vorausgesetzt dass Autor und Quelle genannt werden.


Gliederung

Text

The ability to image the choroid by OCT using enhanced depth imaging or long-wavelength light sources has triggered substantial scientific interest in the role of the choroidal vasculature (CV) in ocular pathology, especially in retinal vascular diseases such as diabetic macular edema (DME). However, the need for specific scanning protocols or equipment, and the laborious and variable process of manual evaluation limit the clinical application of quantitative choroidal analysis. In this study, we investigated the CV using an automated validated segmentation in standard spectral-domain optical coherence tomography (SD-OCT), and correlated the extent of CV pathology with retinal vascular damage.

284 standardized Cirrus SD-OCT 512×128 scans of 142 patients with treatment naïve DME in one eye were analyzed by certified graders of the Vienna Reading Center.After automated detection of the retinal layers, the entire CV was segmented using Hessian analysis-based object detection followed by classic region-growing segmentation to quantify choroidal thickness (CT) by fitting a thin-plate spline on both sides and determining local surface-to-surface distances. CT was calculated after early treatment diabetic retinopathy study grid centering at the fovea and at the peak of edema. Data were compared to fellow eyes without DME and to a healthy population. Furthermore, CT was compared to the maximum area of leakage (LA) measured on standard late-phase fluorescein angiography images.

Mean CT was 175 μm ± 23 μm in DME patients (A), 190 μm ± 23 μm in healthy controls (B) and 177 μm ± 20 μm in non-DME fellow eyes (C) (ANOVA ABC: p=0.03, t-test AC: p=0.59). A trend for a correlation between retinal thickness (RT) and CT at the foveal central millimeter was observed (r2=0.104, p=0.14). No significant correlation between RT and CT at the center or at the peak of edema could be detected. In eyes with DME, the mean LA was 23mm2. A minimal negative correlation was found between LA and CT (r= –0.16, p=0.02). CV segmentation and assessment of CT demonstrated consistent, significant thinning of the choroidal vascular compartment across the entire macula in patients with diabetes regardless of the presence of DME. Further comparison between the retinal and choroidal vascular disease state revealed no significant correlation of the individual degree of vasculopathy.