July 2020
Volume 61, Issue 9
Open Access
ARVO Imaging in the Eye Conference Abstract  |   July 2020
Longitudinal assessments of retinal degeneration after excitotoxic injury using an end-to-end pipeline with deep learning-based automatic layer segmentation
Author Affiliations & Notes
  • Da Ma
    School of Engineering Science, Simon Fraser University, Burnaby, British Columbia, Canada
  • Wenyu Deng
    Department of Ophthalmology, NYU Grossman School of Medicine, NYU Langone Health, New York University, New York, New York, United States
  • Xinlei Wang
    Department of Ophthalmology, NYU Grossman School of Medicine, NYU Langone Health, New York University, New York, New York, United States
  • Sieun Lee
    School of Engineering Science, Simon Fraser University, Burnaby, British Columbia, Canada
    Department of Ophthalmology & Visual Sciences, University of British Columbia, Vancouver, British Columbia, Canada
  • Joanne Matsubara
    Department of Ophthalmology & Visual Sciences, University of British Columbia, Vancouver, British Columbia, Canada
  • Marinko Sarunic
    School of Engineering Science, Simon Fraser University, Burnaby, British Columbia, Canada
  • Mirza Faisal Beg
    School of Engineering Science, Simon Fraser University, Burnaby, British Columbia, Canada
  • Kevin C Chan
    Department of Ophthalmology, NYU Grossman School of Medicine, NYU Langone Health, New York University, New York, New York, United States
    Department of Radiology, NYU Grossman School of Medicine, NYU Langone Health, New York University, New York, New York, United States
  • Footnotes
    Commercial Relationships   Da Ma, None; Wenyu Deng, None; Xinlei Wang, None; Sieun Lee, None; Joanne Matsubara, None; Marinko Sarunic, None; Mirza Faisal Beg, None; Kevin Chan, None
  • Footnotes
    Support  This work was supported in part by the National Institutes of Health R01-EY028125 (Bethesda, Maryland); BrightFocus Foundation G2013077, G2016030 and G20190103 (Clarksburg, Maryland); and an Unrestricted Grant from Research to Prevent Blindness to NYU Langone Health Department of Ophthalmology.
Investigative Ophthalmology & Visual Science July 2020, Vol.61, PB0053. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Da Ma, Wenyu Deng, Xinlei Wang, Sieun Lee, Joanne Matsubara, Marinko Sarunic, Mirza Faisal Beg, Kevin C Chan; Longitudinal assessments of retinal degeneration after excitotoxic injury using an end-to-end pipeline with deep learning-based automatic layer segmentation. Invest. Ophthalmol. Vis. Sci. 2020;61(9):PB0053.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : Upon binding and activation by excitatory neurotransmitters such as glutamate, N-methyl-D-aspartate (NMDA) receptors can trigger Ca2+ influx in the retina for mediating visual signal transfer. However, overstimulation of NMDA receptors may lead to excitotoxic retinal injury, which can manifest as a reduction in retinal thickness. In order to investigate how such retinal thinning progresses across layers during excitotoxicity, we used optical coherence tomography (OCT) and deep learning-based retinal layer segmentation to assess longitudinally the effects of NMDA overstimulation in the eye.

Methods : Animal preparation: Excitotoxic retinal injury was induced in 5 Long Evans rats via a single 2 µL intravitreal injection of 150 nmol NMDA dissolved in 0.9% saline solution into the right eye, with the left eye untreated and serving as an internal control. Spectral-domain OCT (Bioptigen Inc) was used to image a 2.5×2.5×2 mm3 volume centered on the optic nerve head (ONH) for both eyes before and 1, 3, 7, 28 days after NMDA injection.
Layer segmentation and thickness estimation: Five retinal layers were manually segmented on each B-scan of one independent volume that served as training data for automatic segmentation using a variant of U-net with Resnet block. Axial motion was corrected by maximizing the cross-correlation between adjacent B-scans. The thickness for each layer was calculated from the boundary voxels. The retinal thickness within the field of view was measured with the exception of the ONH region. Pairwise T-tests were conducted to compare the retinal thickness of the NMDA experimental eye and the control eye across time. Multiple comparisons were controlled with False discovery rate=0.1.

Results : Fig 1 shows the sample images of automatic retinal layer segmentation (Fig 1A), as well as the thickness measurements of 5 retinal layers for both the injured and control eyes (Fig 1B) before and 28 days after unilateral NMDA injection. Gradual retinal degeneration was observed in both inner and outer layers of the NMDA-injected eyes to different extents (Fig 2).

Conclusions : We have demonstrated an end-to-end pipeline to measure the spatiotemporal profiles of retinal degeneration due to unilateral NMDA injection. The longitudinal consistency for the control eye also demonstrates the robustness of the measurements.

This is a 2020 Imaging in the Eye Conference abstract.

 

 

×
×

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.

×