Ischemic region detection using Optical Coherence Tomography Angiography

Abstract

Purpose : To demonstrate that OCT angiography is able to quantify avascular regions of the retina and to evaluate the performance of the algorithm in both normal and diabetic patients.

Methods : Data was acquired on a CIRRUS™ HD-OCT 5000 with AngioPlex™ OCT Angiography (ZEISS, Dublin, CA)
using a 3mm OCTA scan. The superficial retinal layer angiography en face image was first enhanced using a vessel enhancement filter and binarized. The binary images were inverted and cleared by removing the regions smaller than a certain size. The foveal avascular zone was detected and removed. The total avascular area in one image was calculated through integration of the avascular region map. 15 normal subjects and two DR patients were recruited to participate this study. For each normal subject, the vasculature imaging procedure was performed three times on three machines to test the system repeatability. Then two DR patients were imaged by one of the tested machine to test the system performance on diseased eyes.

Results : Fig.1 shows the results of each step of avascular region detection algorithm. In order to test the system
repeatability, we used the image results before removing the small regions. The coefficient of repeatability observed in 15 normal subjects imaged on 3 machines was 3.2%. Fig.2 demonstrates the avascular region detection comparison between normal and DR subjects. The left two columns are the image results captured from normal subjects. The right column are results captured from DR patients. The top row shows the vasculature image overlaid with avascular region (red color). The middle row demonstrates the avascular regions after remove the FAZ. The bottom row shows the avascular regions after remove the smaller regions. As demonstrated in Fig.2, the DR patients have much larger avascular areas compared to normal subjects.

Conclusions : It is possible to detect avascular regions from OCT angiography images. The avascular map and quantitative results have potential to be a useful for DR diagnosis and monitoring.

This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.

 

Fig.1 Avascular region detection. a) original image; b) enhanced image; c) binary image ; d), e) and f) avascular region with FAZ, without FAZ and remove smaller regions.

 

Fig.2 Avascular region detection of normal (left two columns) and DR (right two columns) subjects. Top: vessel images overlaid with avascular regions (red); Middle: avascular regions after remove FAZ; Bottom: avascular regions after clear smaller ones.