June 2015
Volume 56, Issue 7
ARVO Annual Meeting Abstract  |   June 2015
Automated Segmentation of Reticular Pseudodrusen and Regular Drusen in Eyes with Non-neovascular AMD
Author Affiliations & Notes
  • Zhihong Hu
    Doheny Eye Institute, UCLA, Los Angeles, CA
  • Jun Xiao
    Doheny Eye Institute, UCLA, Los Angeles, CA
    The Second Hospital of Jilin University, Jilin, China
  • Kiran Nandanan
    Doheny Eye Institute, UCLA, Los Angeles, CA
  • Srinivas R Sadda
    Doheny Eye Institute, UCLA, Los Angeles, CA
  • Footnotes
    Commercial Relationships Zhihong Hu, None; Jun Xiao, None; Kiran Nandanan, None; Srinivas Sadda, Allergan (C), Carl Zeiss Meditec (C), Carl Zeiss Meditec (F), Optos (C), Optos (F), Optovue, Inc. (F), Regeneron (C)
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science June 2015, Vol.56, 5156. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Zhihong Hu, Jun Xiao, Kiran Nandanan, Srinivas R Sadda; Automated Segmentation of Reticular Pseudodrusen and Regular Drusen in Eyes with Non-neovascular AMD. Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):5156.

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

To quantify regular drusen (RD) and reticular pseudodrusen (RPD) in optical coherence tomography (OCT) images from eyes with non-neovascular age-related macular degeneration (AMD) using an automated analysis algorithm.


In this IRB-approved study, three groups of subjects underwent spectral domain (SD) OCT imaging in one eye using a macular cube protocol (Spectralis OCT). Group 1 consisted of 17 healthy subjects (referred as "Normal"); group 2 had 6 AMD subjects with only RD (referred as "Non-reticular"); and group 3 included 16 AMD subjects with both RD and RPD (referred as "Reticular"). A graph-based approach was applied to automatically segment 5 sub-retinal boundaries: ellipsoid zone (EZ) outer, interdigitation zone (IZ), retinal pigment epithelium (RPE) inner, RPE outer, and choroid inner (CI). The segmentation was manually corrected as needed by a certified OCT grader. Using these segmented boundaries, three layers were generated for quantitative thickness analysis: (1) RPD layer bounded by IZ and RPE inner, (2) RD layer bounded by RPE outer and CI, and (3) RPE-drusen complex (RPEDC) bounded by EZ and CI. Thicknesses of these layers were computed in a 4mm*4mm grid (including 4 sub-quadrants,) as well as a circular grid (r = 1.44 mm) centered on the fovea. Values were compared among the 3 groups using t-test.


Fig. 1 shows the mean regional thicknesses in RPD and RD layers and Fig. 2 shows these values in RPEDC layer. The mean thickness of RPD layer in 4mm*4mm grid was significantly greater (p < 0.01) than in the circular grid (r = 1.44mm) with the highest mean thickness in the superior-nasal quadrant. Conversely, the mean thicknesses of RD and RPEDC layers in 4mm*4mm grid was significantly smaller than that in the circular region for both "Non-reticular" (p < 0.01) and "Reticular" groups (p < 0.01). When RPD and RD were both present, the mean RD layer thickness was greater than that in the "Non-reticular" group in all the regions, but not statistically significant.


Automated segmentation analysis suggests that RPD are most extensive/thicker in the superior-nasal quadrant of the macula whereas RD are thicker centrally, highlighting differences in regional distribution of these lesions. The combined RPEDC layer shows a similar regional distribution to RD suggesting that this measure is more driven by the presence of RD.  



This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.