Abstract
Purpose :
The National Eye Institute has identified the need to engineer and apply new techniques to study blood flow in the retina and choroid. We propose the use of computer vision and video processing to elucidate the role of the choroid in retinal pathologies that involve abnormal perfusion, such as diabetic choroidopathy. To this goal, we describe an innovative technique by which retinal and choroidal blood flow can be visualized and quantified without the use of contrast dyes or specialized equipment.
Methods :
Preliminary retinal video was obtained from a surgical retina video library (Truevision, Goleta, CA). Videos of different organs were recorded while vessels were occluded via a blood-pressure cuff, using consumer-grade digital video cameras (NEX-5T, a7sii; Sony, New York, NY). All other retinal videos were taken using a fundus camera (50X; Topcon, Oxland, NJ) modified to support the above digital video cameras. Videos were processed and quantified using experimental software (MATLAB; Mathworks, Natick, MA).
Results :
We demonstrate our software enhancement technique to visualize choroidal vasculature from retinal videos. Plotting signal intensity reveals a pulsatile-like waveform. As pressure from the blood-pressure cuff decreases on vasculature, signal intensity and amplitude of the revealed pulsations increases, correlating with increased blood flow. In retinal videos of a healthy subject, this software enhancement enables increased visibility of choroidal vasculature while also having a reproducible quantification (ICC = 0.840, 95% CI = 0.530-0.981).
Conclusions :
We have demonstrated the capability to enhance the visualization of retinal and choroidal vasculature in a reproducible and quantifiable fashion. We initiated a pilot study to further compare this technique with Indocyanine Green angiography, in the hopes of applying this technique clinically.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.