PURPOSE: To compare neovascularization identified in proliferative diabetic retinopathy (PDR) eyes by widefield swept-source (SS) OCT angiography (OCTA) using vitreoretinal interface (VRI) slab images, composed by automated and manual segmentation, with that identified by fluorescein angiography (FA). DESIGN: Retrospective study. PARTICIPANTS: Forty-two eyes of 30 treatment-naïve PDR patients who visited the outpatient clinic of the Department of Ophthalmology, Shinshu University, from June 2018 through October 2019. METHODS: All patients underwent comprehensive ophthalmologic examinations, including SS-OCTA and FA. MAIN OUTCOME MEASURES: Neovascularization detected by en face SS-OCTA 15 × 15-mm VRI slab images and by FA in the same 15 × 15-mm areas were compared in terms of number and structure. RESULTS: Among 100 neovascularizations detected by FA, 73 also were visualized as neovascularization in SS-OCTA VRI slab images using automated segmentation. The sensitivity of VRI slab images for detecting neovascularization was 73%. Among the remaining 27 neovascularizations detected only by FA, but not by VRI slab, 15 were intraretinal microvascular abnormalities with fluorescence leakage, 1 was a diabetic papillopathy, and 11 were flat neovascularizations on the internal limiting membrane surface that were missed because of segmentation error. Conversely, among the 98 neovascularizations detected on VRI slab images, 25 were not detected as neovascularizations by FA. They included 9 small neovascularizations that exhibited too little leakage on FA and 16 false-positive results that were the result of segmentation errors. After reconstruction of SS-OCTA VRI slab images by means of manual segmentation, the sensitivity of VRI slab images for detecting neovascularizations increased to 84%. CONCLUSIONS: The efficacy of SS-OCTA VRI slab images for detecting neovascularizations in PDR was comparable with that of FA. Swept-source OCTA VRI slab images may be better than FA for identifying intraretinal microvascular abnormalities and diabetic papillopathy from neovascularizations. Notably, however, FA and SS-OCTA VRI slab images demonstrated differences in identification efficacy in cases of small and flat neovascularizations. Further exploration of SS-OCTA technology is warranted to address this issue.
PURPOSE: To compare neovascularization identified in proliferative diabetic retinopathy (PDR) eyes by widefield swept-source (SS) OCT angiography (OCTA) using vitreoretinal interface (VRI) slab images, composed by automated and manual segmentation, with that identified by fluorescein angiography (FA). DESIGN: Retrospective study. PARTICIPANTS: Forty-two eyes of 30 treatment-naïve PDR patients who visited the outpatient clinic of the Department of Ophthalmology, Shinshu University, from June 2018 through October 2019. METHODS: All patients underwent comprehensive ophthalmologic examinations, including SS-OCTA and FA. MAIN OUTCOME MEASURES: Neovascularization detected by en face SS-OCTA 15 × 15-mm VRI slab images and by FA in the same 15 × 15-mm areas were compared in terms of number and structure. RESULTS: Among 100 neovascularizations detected by FA, 73 also were visualized as neovascularization in SS-OCTA VRI slab images using automated segmentation. The sensitivity of VRI slab images for detecting neovascularization was 73%. Among the remaining 27 neovascularizations detected only by FA, but not by VRI slab, 15 were intraretinal microvascular abnormalities with fluorescence leakage, 1 was a diabetic papillopathy, and 11 were flat neovascularizations on the internal limiting membrane surface that were missed because of segmentation error. Conversely, among the 98 neovascularizations detected on VRI slab images, 25 were not detected as neovascularizations by FA. They included 9 small neovascularizations that exhibited too little leakage on FA and 16 false-positive results that were the result of segmentation errors. After reconstruction of SS-OCTA VRI slab images by means of manual segmentation, the sensitivity of VRI slab images for detecting neovascularizations increased to 84%. CONCLUSIONS: The efficacy of SS-OCTA VRI slab images for detecting neovascularizations in PDR was comparable with that of FA. Swept-source OCTA VRI slab images may be better than FA for identifying intraretinal microvascular abnormalities and diabetic papillopathy from neovascularizations. Notably, however, FA and SS-OCTA VRI slab images demonstrated differences in identification efficacy in cases of small and flat neovascularizations. Further exploration of SS-OCTA technology is warranted to address this issue.
Authors: Hasenin Al-Khersan; Jonathan F Russell; Thomas A Lazzarini; Nathan L Scott; John W Hinkle; Nimesh A Patel; Nicolas A Yannuzzi; Benjamin J Fowler; Rehan M Hussain; Anita Barikian; Jayanth Sridhar; Stephen R Russell; Luis J Haddock; William E Smiddy; Seenu M Hariprasad; Yingying Shi; Liang Wang; William Feuer; Giovanni Gregori; Philip J Rosenfeld Journal: Am J Ophthalmol Date: 2020-12-11 Impact factor: 5.258