Investigative Ophthalmology & Visual Science Cover Image for Volume 65, Issue 9
July 2024
Volume 65, Issue 9
Open Access
ARVO Imaging in the Eye Conference Abstract  |   July 2024
Automated Outer Retinal Thickness Measurement in Swept-Source OCT imaging using contrast-enhanced OAC images for supervised deep-learning based segmentation
Author Affiliations & Notes
  • Bhagavath Sivathanu Kumar
    Bioengineering, University of Washington, Seattle, Washington, United States
  • Ruikang K. Wang
    Bioengineering, University of Washington, Seattle, Washington, United States
  • Yuxuan Cheng
    Bioengineering, University of Washington, Seattle, Washington, United States
  • Footnotes
    Commercial Relationships   Bhagavath Sivathanu Kumar, None; Ruikang Wang, Carl Zeiss Meditec (C), Carl Zeiss Meditec, Colgate Palmolive Company, Estee Lauder Inc (F); Yuxuan Cheng, None
  • Footnotes
    Support  None
Investigative Ophthalmology & Visual Science July 2024, Vol.65, PB0023. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Bhagavath Sivathanu Kumar, Ruikang K. Wang, Yuxuan Cheng; Automated Outer Retinal Thickness Measurement in Swept-Source OCT imaging using contrast-enhanced OAC images for supervised deep-learning based segmentation. Invest. Ophthalmol. Vis. Sci. 2024;65(9):PB0023.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : The progression of the ORL thickness in patients has been previously identified as a suitable biomarker for tracking Age-related Macular degeneration (AMD) progression. This study aims to establish an automated algorithm to measure the thickness of the Outer Retinal Layer (ORL) of an SS-OCT scan, with the aid of contrast enhanced OAC (Optical Attenuation Correction) depth-resolved images as an aide to the supervised deep learning based segmentation.

Methods : The accuracy of the UNet-based segmentation algorithm was improved by incorporating contrast-enhanced OAC depth-resolved images into the training dataset. This improvement was shown through a comparative study with a base dataset consisting of 16 SS-OCT scans. The study evaluated performance using metrics such as IoU, Dice coefficient, and Mean Squared Error from a validation dataset. It generated RPE and OPL segmentation lines, measuring ORL thickness between these lines. Analysis of ORL thickness across 1mm, 3mm, and 5mm fovea-centered circles in AMD stages was done, noting Drusen presence in scans. Additionally, the drusen volume calculations took Hyper Transmission defects into consideration by overlaying them as a negative mask on thickness maps.

Results : An improvement in the performance metrics was observed between the depth-resolved OAC training dataset and a normal SS-OCT B-scan dataset due to the enhanced contrast of the RPE with the OAC images. An average increase in the performance metrics was observed, with the IoU increasing from 0.9305 to 0.9506, the Dice coefficient from 0.9538 to 0.9693, and the MSE decreasing from 0.0032 to 0.0024. Strong correlations of the algorithm between automatic segmentation and manual segmentation are noted (Mean bias in 3mm circle = 0.50 μm and 5mm circle = 1.23 μm). Comparative analysis of ORL thickness for the same AMD patient at different time periods shows changes in ORL thickness in both the 3mm and 5mm circles.

Conclusions : The performance of the Unet algorithm to automatically segment and measure ORL thickness was significantly improved with the use of depth-resolved OAC images. The analysis of the obtained thickness measurements reveals significant differences in ORL thickness within the same patient across different stages of AMD treatment, indicating its potential to be used as a biomarker for assessing AMD progression.

This abstract was presented at the 2024 ARVO Imaging in the Eye Conference, held in Seattle, WA, May 4, 2024.

 

 

×
×

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.

×