A simulation study of fluorescein attenuation and dynamic tracer kinetic model-extracted retinal blood flow from human retinal fluorescein videoangiographies

Diabetes mellitus catalyzes blood-retinal barrier degeneration and alterations to retinal hemodynamics, leading to changes in retinal vascular permeability and retinal blood flow (F), ultimately resulting in diabetic retinopathy (DR). DR currently lacks preclinical quantitative biomarkers capable of detecting changes in retinal vasculature, which occur well before the presentation of the currently used qualitative clinical markers of DR. The objective of this work is to display that fluorescein signal attenuation in retinal arteries compared to retinal tissue creates overestimation of F. F is extracted from fluorescein videoangiographies of the human retina using dynamic tracer kinetic modeling (DTKM). In seven healthy control subjects, twenty-two subjects with diabetes mellitus and no clinical signs of retinopathy (DMnoDR), and seven subjects with mild non-proliferative diabetic retinopathy (NPDR), the DTKM-extracted F was quantitatively estimated at values 10-20 times higher than previous human studies. These overestimations in retinal blood flow align with the fact that, when assuming Beer-Lambert Law attenuation, fluorescein attenuation in blood is expected to be 2-3 orders-of-magnitude greater than in tissue when considering fluorescein’s peak emission wavelength (~520nm). Using Beer Lambert Law and optical properties at 570 nm, the ratio of retinal tissue and blood’s detected fluorescein intensities at pathlengths equivalent to retinal tissue thicknesses, 170 μm and 267 μm, and retinal artery diameters, 79.0 μm and 111.2 μm, were found to be between ~12-17. The increased retinal arterial diameter had a negligible effect on the ratio of retinal tissue and blood’s detected fluorescein intensities, whereas the increased retinal thickness increased this ratio. A Monte Carlo simulation study will be conducted to provide further proof.