Quantification of intermittent retinal capillary perfusion in retinal vein occlusion and proliferative diabetic retinopathy.

Abstract

Objective: To detect and quantify intermittent capillary perfusion using optical coherence tomography angiography (OCTA) in patients with branch retinal vein occlusion (BRVO), central retinal vein occlusion (CRVO), proliferative diabetic retinopathy (PDR), and healthy control eyes.

Methods: OCTA images were acquired from patients with BRVO(n = 9), CRVO(n = 8), PDR(n = 8) and healthy controls(n = 10). Five 6 × 6 mm scans were registered and averaged at baseline (T0) and thirty minutes after (T30) into single en-face images of the superficial and deep vascular complexes (SVC and DVC). Pixels were labeled as vessel or non-vessel using a previously published machine learning model. Loss of Perfusion (LoP) was defined as the percentage of vessel pixels present in T0 image that disappeared at T30, and Gain of Perfusion (GoP) was defined as the percentage of vessel pixels that appeared in T30 image. The amount of intermittent capillary perfusion was the sum of GoPLoP.

Results: Patients with PDR, CRVO and BRVO showed significantly higher GoPLoP values than controls in both the macular and temporal regions. The temporal region generally exhibited significantly greater GoPLoP values than the macular region. Layer analysis indicated a significantly higher GoPLoP within the DVC compared to the SVC. There was a significant negative correlation between perfusion density and perfusion variability.

Conclusion: Our results demonstrate higher GoPLoP in BRVO, CRVO, and PDR patients compared to controls. This measure may be utilized as a novel biomarker of tissue hypoxia. Further studies are necessary to better elucidate the role of GoPLoP in monitoring disease progression and treatment efficacy of retinal vascular diseases.