Abstract
Purpose :
Retinal vessel calibers as extracted from fundus photos are associated with ocular and systemic diseases. We investigated whether total retinal blood flow can be modeled from fundus photos. This approach consisted of two steps: 1. Modeling of retinal blood flow according to Murray’s law based on Doppler OCT 2. Apply these findings to fundus photos obtained from the Singapore Chinese Eye Study (SCES, n=706) and investigate associated factors.
Methods :
The retinal blood flow measurements were performed in 40 subjects. OCT phase images were used to measure blood velocities and vessel calibers in arteries (vAi,, dAi) and veins (vVi,, dVi). The blood flow in each vessel was calculated as (QA,i; Qv,j) and total retinal blood flow was calculated as sum of all QA,i for arteries and sum of all QV,j for veins. Log-log plots between dAi and QAi and dVj and QVj were drawn. Using correlation analysis qV and qA were calculated from the equations log QVj = qV x logVj + xV and log QAj = qA x logAj + xA. According to Murray’s law a value of 3 is expected for both qV and qA..Retinal vessel calibers from SCES were measured using a semi-automated program (SIVA). According to the law of mass conservation QA,tot = QV,tot. Assuming one value of qV = 3 for all venules in a particular subject allows for calculation of qA even when the blood velocities are unknown.
Results :
There was good agreement between QA,tot = 41.2 ± 6.4 μl/min and QV,tot= 41.9 ± 6.6 μl/min. The Log-Log graphs between dAi and QAj as well as dVj and QVj revealed a high degree of association (r > 0.97, p < 0.001) with values of qV and qA of 2.97 and 2.99, respectively. The value as obtained for qA in the SCES cohort was 3.01 ± 0.08 at a fixed qV of 3. Multivariate analysis showed that qA was associated with mean blood pressure (β= 0.011, p = 0.007) and QV,tot was negatively associated with age (β= -0.231, p = 0.003).
Conclusions :
We present an approach to model retinal blood flow based on vessel caliber extraction from fundus photos. This is based on Murray’s law and its experimental verification using Doppler OCT. In the present study we observed a dependence of qA on blood pressure, reflecting that qA is the higher the lower the arterio-venous ratio. In addition we show a decline of retinal blood flow with age, which has been reported in previous studies. The present algorithm may be a simple approach to study retinal blood flow abnormalities based on fundus photos.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.