July 2018
Volume 59, Issue 9
Open Access
ARVO Annual Meeting Abstract  |   July 2018
ReVAS: An open-source tool for eye motion extraction from retinal videos obtained with scanning laser ophthalmoscopy
Author Affiliations & Notes
  • Mehmet N Agaoglu
    School of Optometry, University of California, Berkeley, Berkeley, California, United States
  • Matthew Sit
    School of Optometry, University of California, Berkeley, Berkeley, California, United States
  • Derek Wan
    School of Optometry, University of California, Berkeley, Berkeley, California, United States
  • Susana T L Chung
    School of Optometry, University of California, Berkeley, Berkeley, California, United States
  • Footnotes
    Commercial Relationships   Mehmet Agaoglu, None; Matthew Sit, None; Derek Wan, None; Susana Chung, None
  • Footnotes
    Support  NIH Grant R01-EY012810
Investigative Ophthalmology & Visual Science July 2018, Vol.59, 2161. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Mehmet N Agaoglu, Matthew Sit, Derek Wan, Susana T L Chung; ReVAS: An open-source tool for eye motion extraction from retinal videos obtained with scanning laser ophthalmoscopy. Invest. Ophthalmol. Vis. Sci. 2018;59(9):2161.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : Our eyes are never stable even when we maintain our gaze steadily. The variability and characteristics of eye motion during fixation have been proposed as diagnostic biomarkers for certain neurological and visual disorders. Scanning laser ophthalmoscopes (SLOs) are used in clinical and research settings to study fixational eye movements (FEMs). However, their use is limited to estimates of fixation stability due to their low frame rates (≤30Hz). Having eye motion at a millisecond resolution enables more comprehensive analyses of FEMs. We developed an open-source tool for extracting eye motion at high sampling rates from retinal videos obtained by SLOs.

Methods : The pipeline of the Retinal Video Analysis Suite (ReVAS) consists of (1) pre-processing, (2) strip analysis, and (3) post-processing. In (1), video frames are trimmed, gamma-corrected, and bandpass filtered to remove noise in the images. Since rows within each video frame are scanned at different times, eye motion is embedded within each frame. In (2), each strip (1 or more consecutive rows) of a video frame is cross-correlated with a reference frame to estimate eye motion. Initially, a frame extracted directly from the video is used, and a new reference frame is constructed using the estimated eye motion. Repeating (2) multiple times results in virtually motionless reference frames and improves the accuracy and precision of eye motion estimates. In (3), eye motion data can be filtered, re-referenced onto another reference frame, and classified as (micro)saccades and drifts. We tested ReVAS with >1K videos recorded using a Rodenstock SLO, an eye-tracking SLO, and an adaptive optics SLO from people with normal and abnormal FEMs (amblyopia, macular degeneration).

Results : ReVAS extracts eye motion at any rate from the native sampling rate of raw videos up to the horizontal scan rate of SLOs (>10KHz), and from videos capturing from 0.65 to 1200deg2 retinal areas. Its performance depends on the clarity and abundance of retinal features. A 1s video captured at 30Hz can be analyzed in <2min to get eye motion sampled at 1KHz.

Conclusions : ReVAS facilitates the extraction of eye motion from retinal videos. We expect that its usage could lead to standardized, rich, and reliable datasets of eye motion, which may, in turn, enable using FEMs as biomarkers for early diagnosis of certain disorders.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.

 

Graphical user interface of ReVAS.

Graphical user interface of ReVAS.

×
×

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.

×