July 2019
Volume 60, Issue 9
Open Access
ARVO Annual Meeting Abstract  |   July 2019
Identifying transient flow phenomena in retinal capillaries
Author Affiliations & Notes
  • Phillip A Bedggood
    Optometry & Vision Sciences, University of Melbourne, Parkville, Victoria, Australia
  • Andrew Metha
    Optometry & Vision Sciences, University of Melbourne, Parkville, Victoria, Australia
  • Footnotes
    Commercial Relationships   Phillip Bedggood, None; Andrew Metha, None
  • Footnotes
    Support  Australian Research Council Discovery Project DP180103393
Investigative Ophthalmology & Visual Science July 2019, Vol.60, 4588. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Phillip A Bedggood, Andrew Metha; Identifying transient flow phenomena in retinal capillaries. Invest. Ophthalmol. Vis. Sci. 2019;60(9):4588.

      Download citation file:

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

  • Supplements

Purpose : Regular capillary flow is punctuated by brief but interesting events including white cells, red cell aggregates, low haematocrit, and slow flow. Such events cannot be readily differentiated by velocity measures alone, however, they transiently stabilize capillary appearance. We aimed to quantify this property and determine an appropriate threshold to identify transient flow phenomena.

Methods : Flood illuminated adaptive optics was used to image retinal capillaries at 300 fps in 2 healthy subjects (age 22-23 years), at 750 nm over 3 sec. Regions imaged were 1.25° wide and within 1-3° of fixation. Vessels were automatically segmented from the motion contrast image, and analysed if diameter was <10 µm, length was >20 µm, and illumination lasted >1s (N=70). Measurements were made over 100 ms epochs for each pixel, and averaged by vessel segment. Velocity was determined with a new algorithm which compares the degree of correlation between all pairs of pixels; stability was quantified from Fourier analysis of the autocorrelation function, which was converted into a “decay time” for each pixel, reported as (mean ± standard deviation).

Results : The decay time image shows bright, localised regions which transiently appear and propagate across the field. Such events look very different from background tissue and the vessel network at large, which are typically dark (decay time 14.5±5.0 ms and 15.0±4.0 ms respectively). To flag these events we adopted outlier analysis, based on medians, which suggested a threshold of 22.8 ms to differentiate high decay time (26.6±4.1 ms, 3.5% of data) from normal flow (14.2±2.9 ms). However, a fixed threshold does not allow for natural variation in decay time as velocity varies between segments and across the cardiac cycle. Outlier analysis on individual vessels mitigated this factor, returning thresholds from 14.0 to 30.9 ms (1.1 to 12.2 ms above their means), where fast thresholds corresponded to fast vessels (p<1e-6, R2=0.59). Abnormal events were flagged in 2.2% of observations, with manual inspection confirming that every event flagged corresponded to either unusually slow flow or extended lengths of plasma and/or cells.

Conclusions : We have developed a metric of capillary blood column stability which can identify transient departures from typical flow patterns in the healthy retina.

This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.


Figure 1. Example data showing motion contrast (top) and, in a 100 ms epoch, decay time (middle) and velocity (bottom).

Figure 1. Example data showing motion contrast (top) and, in a 100 ms epoch, decay time (middle) and velocity (bottom).


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.