June 2023
Volume 64, Issue 8
Open Access
ARVO Annual Meeting Abstract  |   June 2023
Human Expert Grading versus Automated Quantification of Fluid Volumes in nAMD, DME and BRVO
Author Affiliations & Notes
  • Felix Goldbach
    Ophthalmology and Optometry, Medizinische Universitat Wien, Wien, Wien, Austria
  • Bianca S Gerendas
    Ophthalmology and Optometry, Medizinische Universitat Wien, Wien, Wien, Austria
  • Oliver Leingang
    Ophthalmology and Optometry, Medizinische Universitat Wien, Wien, Wien, Austria
  • Thomas Alten
    Ophthalmology and Optometry, Medizinische Universitat Wien, Wien, Wien, Austria
  • Jonas Brugger
    Center for Medical Statistics, Informatics, and Intelligent Systems (CeMSIIS), Medizinische Universitat Wien, Wien, Wien, Austria
  • Hrvoje Bogunovic
    Ophthalmology and Optometry, Medizinische Universitat Wien, Wien, Wien, Austria
  • Ursula Schmidt-Erfurth
    Ophthalmology and Optometry, Medizinische Universitat Wien, Wien, Wien, Austria
  • Footnotes
    Commercial Relationships   Felix Goldbach None; Bianca S Gerendas Roche, Code C (Consultant/Contractor), Novartis, Code C (Consultant/Contractor), Bayer, Code C (Consultant/Contractor), Zeiss, Code C (Consultant/Contractor), DXS, Code F (Financial Support); Oliver Leingang None; Thomas Alten None; Jonas Brugger None; Hrvoje Bogunovic Apellis, Code F (Financial Support), Heidelberg Engineering, Code F (Financial Support); Ursula Schmidt-Erfurth Apellis, Code C (Consultant/Contractor), Novartis, Code F (Financial Support), Genentech, Code F (Financial Support), Apellis, Code F (Financial Support), Kodiak, Code F (Financial Support), RetInSight, Code F (Financial Support), RetInSight, Code P (Patent)
  • Footnotes
    Support  None
Investigative Ophthalmology & Visual Science June 2023, Vol.64, 1286. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Felix Goldbach, Bianca S Gerendas, Oliver Leingang, Thomas Alten, Jonas Brugger, Hrvoje Bogunovic, Ursula Schmidt-Erfurth; Human Expert Grading versus Automated Quantification of Fluid Volumes in nAMD, DME and BRVO. Invest. Ophthalmol. Vis. Sci. 2023;64(8):1286.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : To compare a validated automated deep learning algorithm with trained and experienced human graders from the Vienna Reading Center (VRC) in identifying intra- (IRF) and subretinal fluid (SRF) in OCT scans of patients treated for neovascular age-related macular degeneration (nAMD), diabetic macular edema (DME) and branch retinal vein occlusion (BRVO).

Methods : This study was a posthoc analysis of multicenter clinical trial data from the VRC imaging database. Imaging data was prospectively collected during randomized multicenter clinical trials. OCT scans were analyzed using the Vienna Fluid Monitor Version 2 (RetInSight, Vienna, Austria) to compute IRF and SRF volumes in the central 1mm (CMM), 3mm and 6mm disc areas and the parafoveal and perifoveal ETDRS ring region and the region outside of the standard ETDRS grid. These fluid volumes were compared to fluid presence graded by trained and experienced graders of the VRC. Area under the curve (AUC) of receiver operating characteristic was applied as a measure of concordance.

Results : 6898 OCT scans were analyzed for fluid volumes and presence of IRF and SRF. For nAMD/DME /BRVO in the CMM: overall concordance for detection of IRF and SRF between the algorithm and manual grading reached an AUC of 0.94/0.92/0.98 and 0.89/0.95/0.92, respectively. If IRF and/or SRF was graded to be present in the CMM, automated quantification determined a mean fluid volume of 8.59/39.66/106.08 (nL) and 8.18/13.14/36.04 (nL) respectively; and 0/0/0.06 (nL) and 0/0/0.34 (nL) respectively if no fluid was graded to be present. When applying a cutoff point of 1/5/10 nanoliters for fluid volume results of IRF and SRF in the CMM (determined by the algorithm) consensus with expert grading in terms of identifying fluid presence/absence was reached in 90/88/86 (%) and 86/86/83 (%) respectively of the cases. Results from the other regions were analyzed in the same manner.

Conclusions : Detection of central, peri- and parafoveal IRF and SRF using a deep learning approach showed high concordance with human grading across different retinal fluid-associated diseases. In addition to information about presence of fluid, automated quantification of retinal fluid provides volumetric information, and its fluid detection is comparable to human detection. Thus, automated fluid quantification is a feasible tool for objective treatment decision support and disease monitoring in clinical practice.

This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.

×
×

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.

×