April 2014
Volume 55, Issue 13
ARVO Annual Meeting Abstract  |   April 2014
Automatic Quantification of RPE Layer Thickness Using a Gaussian Curve Fitting Method in OCT
Author Affiliations & Notes
  • Shu-Wei Sun
    Loma Linda University, Loma Linda, CA
    University of California, Riverside, CA
  • Chen-Fang Chung
    Loma Linda University, Loma Linda, CA
  • Hsiao-Fang Liang
    Loma Linda University, Loma Linda, CA
  • Footnotes
    Commercial Relationships Shu-Wei Sun, None; Chen-Fang Chung, None; Hsiao-Fang Liang, None
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science April 2014, Vol.55, 3840. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Shu-Wei Sun, Chen-Fang Chung, Hsiao-Fang Liang; Automatic Quantification of RPE Layer Thickness Using a Gaussian Curve Fitting Method in OCT. Invest. Ophthalmol. Vis. Sci. 2014;55(13):3840.

      Download citation file:

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

  • Supplements

Optical coherence tomography (OCT) is a powerful tool for retina disease diagnosis and research. For Age-related Macular Degeneration (AMD), OCT provides early detection of morphological alterations in the Retinal Pigment Epithelium (RPE) layer. Automating the quantification of RPE on OCT will significantly improve the AMD diagnosis. Here, we proposed an automatic procedure, incorporating a Gaussian curve fitting, to quantify the thickness of retinal pigment epithelium (RPE) layer of mouse retina tissue in vivo.


Six female 8-week-old C57BL/6 healthy mice were used. For OCT scans, animals were anesthetized with 1.5% Isoflurane/oxygen. The core body temperature was maintained with a warm water circulating pad. The mouse pupil was dilated with 1% tropicamide (Bausch & Lomb Inc.), followed by artificial tears for preventing corneal desiccation. The mouse eye was imaged with a Envisu 2200-HR SD-OCT imaging system (Bioptigen, Durham, NC). Data was acquired with a rectangular volume scan with area of 1.6 mm by 1.6 mm (1000 a-scans/b-scan, 3 frames/b-scan, total 100 b-scans). For data analysis, OCT images were transferred to a Windows-Based computer. Volumetric data (1000x1024x100) were imported and noise was reduced via a 3-voxel smoother followed by retinal straightening. The relative location of RPE was detected using a combination of signal intensity and its distance relative to the retinal edge. Gaussian cure was then applied. The width and peak location of the Gaussian were used for the estimation of RPE thickness and location, respectively.


We superimposed the quantified RPE layer on the B-scans, and found that the distribution of RPE was fairly accurate (Fig. 1). The RPE thickness was quantified ~44 µm, which is similar to previous reports. It was worth to note that the blood vessels on the surface of retina caused shadows across the retina. As such the signals of RPE were dramatically reduced. Our method showed the RPE quantification was not affected by shadows; the regions with and without shadows showed similar results with no significant difference (Fig. 2).


We proposed a computer-aided automatic procedure for RPE quantification. The use of Gaussian curve fitting method was demonstrated to accurately quantify the RPE thickness. Our method overcame the shadow-effects on OCT to provide an accurate RPE thickness quantification of the entire retina.

Keywords: 701 retinal pigment epithelium • 551 imaging/image analysis: non-clinical • 549 image processing  

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.