Investigative Ophthalmology & Visual Science Cover Image for Volume 65, Issue 7
June 2024
Volume 65, Issue 7
Open Access
ARVO Annual Meeting Abstract  |   June 2024
Reliability of retinal layer thickness measurements in HCs and PwMS from a DL-based segmentation algorithm
Author Affiliations & Notes
  • JI HEE KIM
    JuneBrain Inc., Maryland, United States
    Johns Hopkins University, Baltimore, Maryland, United States
  • Wande Ajose
    JuneBrain Inc., Maryland, United States
  • Yufan He
    Johns Hopkins University, Baltimore, Maryland, United States
  • Omar Ezzedin
    Johns Hopkins University, Baltimore, Maryland, United States
  • Jerry Prince
    Johns Hopkins University, Baltimore, Maryland, United States
  • Shiv Saidha
    Johns Hopkins University, Baltimore, Maryland, United States
  • Samantha Scott
    JuneBrain Inc., Maryland, United States
  • Footnotes
    Commercial Relationships   JI HEE KIM JuneBrain Inc., Code E (Employment); Wande Ajose JuneBrain Inc., Code E (Employment); Yufan He None; Omar Ezzedin None; Jerry Prince JuneBrain Inc., Code F (Financial Support); Shiv Saidha JuneBrain Inc., Code C (Consultant/Contractor), JuneBrain Inc., Code S (non-remunerative); Samantha Scott JuneBrain Inc., Code E (Employment), JuneBrain Inc., Code I (Personal Financial Interest), JuneBrain Inc., Code O (Owner), JuneBrain Inc., Code P (Patent), JuneBrain Inc., Code S (non-remunerative)
  • Footnotes
    Support  NSF STTR Grant 2053315
Investigative Ophthalmology & Visual Science June 2024, Vol.65, 2405. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      JI HEE KIM, Wande Ajose, Yufan He, Omar Ezzedin, Jerry Prince, Shiv Saidha, Samantha Scott; Reliability of retinal layer thickness measurements in HCs and PwMS from a DL-based segmentation algorithm. Invest. Ophthalmol. Vis. Sci. 2024;65(7):2405.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : To evaluate the reproducibility and repeatability of the retinal layer thickness measured by JuneBrain’s (JB’s) algorithm with OCT scans acquired from two different spectral-domain (SD) OCT devices

Methods : For a reproducibility assessment, macular OCT volumes were obtained from the eyes of 17 healthy controls (HCs) and 37 people with multiple sclerosis (PwMS) on the same day using Cirrus HD-OCT and Spectralis SD-OCT devices. Image data were segmented by JB’s deep learning (DL)-based segmentation algorithm. The degree of agreement in thickness measurements of the retinal layers (Fig 1) was quantified using intraclass correlation (ICC) and Bland-Altman analysis. For a repeatability assessment, macular OCT volumes from the eyes of 34 HCs using a Cirrus HD-OCT on two occasions on the same day were segmented by JB’s DL-based segmentation algorithm. Two-way random effect analysis of variance (ANOVA), repeatability coefficient (RC), and ICC were used to evaluate the repeatable thicknesses across the imaging sessions (ISs).

Results : The ICCs for reproducibility of thickness measures from both SD-OCT devices were greater than 0.9 for GCIPL, ONL+PR, and the total retina, while the ICCs of RNFL, INL+OPL, and RPE layers ranged from 0.132 to 0.775. Low mean differences (-1.29 to 0.42 µm) and narrow limits of agreement (LOA) were observed for GCIPL, INL+OPL, and ONL+PR thicknesses. The thicknesses of all retinal layers (Fig 1) across two ISs were not different (p ≥ 0.863). Low RCs (0.513 to 2.994 µm) and high ICCs (0.892 to 0.994) reveal high repeatability.

Conclusions : Thickness measures for retina layers implicated in MS progression showed good agreement between the scanners at the cohort level, despite greater variability at the individual levels. All retinal layer thickness measures are repeatable in two ISs. The reproducible and repeatable thickness measures demonstrate the potential of JB’s DL-based segmentation algorithm to quantify and evaluate clinical parameters in OCT images from PwMS. The variability of LOA suggests that PwMS should use the same OCT device.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.

 

Figure 1. Retinal layers and boundaries in OCT scan. Inner Limiting Membrane (ILM), Retinal Nerve Fiber Layer (RNFL), Ganglion Cell Layer + Inner Plexiform Layer (GCIPL), Inner Nuclear Layer (INL), Outer Plexiform Layer (OPL), Outer Plexiform Layer (ONL), Photoreceptor (PR), Retinal Pigment Epithelium (RPE), and Bruch’s Membrane (BM).

Figure 1. Retinal layers and boundaries in OCT scan. Inner Limiting Membrane (ILM), Retinal Nerve Fiber Layer (RNFL), Ganglion Cell Layer + Inner Plexiform Layer (GCIPL), Inner Nuclear Layer (INL), Outer Plexiform Layer (OPL), Outer Plexiform Layer (ONL), Photoreceptor (PR), Retinal Pigment Epithelium (RPE), and Bruch’s Membrane (BM).

×
×

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.

×