Purchase this article with an account.
Steven T Bailey, Rachel Patel, JIE WANG, Andreas Lauer, J. Peter Campbell, Lee Kiang, Christina J Flaxel, Thomas S Hwang, David Huang, Yali Jia; Projection-resolved optical coherence tomography angiography of choroidal neovascularization. Invest. Ophthalmol. Vis. Sci. 2018;59(9):2620.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To evaluate if projection-resolved optical coherence tomography angiography (PR-OCTA) algorithm reduces projection artifacts with less attenuation of choroidal noevascularization (CNV) flow signal compared to the slab subtraction (SS) technique. Cross-sectional PR-OCTA is compared to conventional OCTA in categorizing CNV as either type 1, type 2 or combined type; using fluorescein angiography (FA) with OCT as the gold standard.
Eyes with subfoveal treatment-naïve CNV were enrolled in a prospective study. Two 3X3 mm OCTA scans (AngioVue, Optovue, Inc) were acquired in each eye and exported for custom processing including manual segmentation when needed and then artifact reduction with either SS or PR-OCTA algorithm. A masked grader encircled the CNV in the outer retinal slab. Then a cluster classification algorithm automatically separated flow pixels (CNV) from nonvascular pixels in the encircled volume and calculated CNV vascular area by summing flow pixels. CNV vascular connectivity was determined by skeletonized CNV length with the requirement of a minimum of five connected pixels. An expert grader (STB) classified CNV as type 1, type 2 or combined type based on gold standard using FA/OCT. Two masked retina specialists (PC and LK) independently classified CNV using conventional cross-sectional OCTA and cross-sectional PR-OCTA.
Seventeen study eyes were enrolled. PR-OCTA preserved CNV pixels better than SS (Fig. 1) which translated to significantly (p=0.018) larger CNV vessel area and length measured by PR-OCTA (0.65±0.61 mm2; 93.2±5.6 mm; mean±SD) compared to SS (0.51±0.50 mm2; 88.3±10.5). Within-visit repeatability of PR-OCTA for CNV vessel area and length was 8.3% and 3.6% (coefficient of variation), compared to 9.9% and 4.4% for SS. In the classification of CNV as type 1, type 2, or combined type, PR-OCTA agreed with the FA standard 88.2%/76.5% (graders 1/2), while conventional OCTA agreed 58.5% /70.6% of the time. The difference was significant for grader 1 (P= 0.025 McNemar’s test) but not for grader 2.
PR-OCTA measurements produced larger CNV vessel area and length, as well as better repeatability, compared to SS – suggesting that PR-OCTA is a superior technique for preserving CNV flow signal. PR-OCTA removes projection artifact on cross-sectional OCTA, which improves the accuracy of classifying CNV.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.
This PDF is available to Subscribers Only