Identification and Characterization of Imaging Technique Errors and Artifacts Using Anterior-Segment OCT for Irido-Corneal Angle Evaluations in Glaucoma

Ophthalmol Glaucoma. 2019 May-Jun;2(3):136-144. doi: 10.1016/j.ogla.2019.02.006. Epub 2019 Feb 13.

Abstract

Purpose: Anterior-segment OCT (AS-OCT) platforms provide noninvasive and quantifiable evaluations of anterior chamber angle (ACA) anatomy. Although these ACA measurements have been validated with high repeatability and reproducibility, there is a paucity of data previously published regarding the technique errors and artifacts (TEA) that may occur during imaging of the ACA. The purpose of this study was to report on the types and frequency of imaging TEA that are seen in ACA evaluation of a large cohort of patients with open-angle glaucoma.

Design: Retrospective, cross-sectional design.

Participants: Images were obtained with AS-OCT from 5082 eyes of 2541 patients with glaucoma and were reviewed at the Doheny Image Reading Center (DIRC).

Methods: Images had been captured using AS-OCT devices from 4 different manufacturers (Zeiss Cirrus [Oberkochen, Germany], Heidelberg Spectralis [Heidelberg, Germany], Optovue RTVue [Fremont, CA], or Zeiss Visante). All images were analyzed by 2 independent and certified DIRC image readers for angle opening distance under Schwalbe's line (SL-AOD), and then re-reviewed by a DIRC-certified principal investigator to adjudicate nonconsensus measurements. Images were defined to have TEA when image readers were unable to confidently measure the SL-AOD.

Main outcome measures: Type and frequency of TEA in the AS-OCT image.

Results: Of 5082 eyes reviewed, 208 images (4.09%) were found to have TEA that restricted proper visualization of irido-corneal angle structures and measurements of SL-AOD. The most common TEA was that the image readers were unable to identify Schwalbe's line (74% of TEA images). Other common TEA included poor visualization of the anterior iris surface (15%), misaligned scan location (11%), irregular corneal irregularities (5%), and illumination-related pupillary abnormalities (4%).

Conclusions: This large-scale study presents the types and frequency of TEA found in AS-OCT-derived angle images. More than 95% of the images reviewed were usable for the quantification of irido-corneal angle measurements. Knowing the common types of TEA may help in enhancing the training of both human image readers and in the design of automated algorithms to improve image acquisition and image interpretation parameters. This will become especially useful as AS-OCT becomes more widely used in clinical practice for anterior chamber assessment.

MeSH terms

  • Anterior Chamber / diagnostic imaging*
  • Artifacts*
  • Cross-Sectional Studies
  • Diagnostic Errors
  • Glaucoma, Open-Angle / diagnosis*
  • Glaucoma, Open-Angle / physiopathology
  • Humans
  • Intraocular Pressure / physiology
  • Reproducibility of Results
  • Retrospective Studies
  • Tomography, Optical Coherence / methods*