Abstract
Purpose :
To compare retinal blood flow velocity (BFV) measurements between the retinal function imager (RFI) and the XyCAM ID, which uses laser speckle contrast imaging, in an in-vitro blood flow model and healthy subjects.
Methods :
A custom in-vitro model was designed with features comparable to the normal human eye. The back surface contained a tube connected to a calibrated syringe pump controlling the flow of human whole blood. The imagers simultaneously captured the model set at a range of known velocities. Ten eyes from ten healthy subjects were also imaged using both devices, and vessels were chosen from the optic disk and macula using reference fundus images (Figure 1). BFVs were computed from the in-vitro setup, arterioles and venules in the RFI using backscattered light from erythrocyte movement by tracing these vessels. The average, standard deviation (SD), and coefficient of variation (CV) of the BFVs from the RFI-generated data were computed. All vessels with CV>0.45 were considered unreliable and excluded from the analysis. BFV from the model and corresponding vessels were computed from the XyCAM ID by analyzing the blurring of laser speckles to obtain BFV over 6 seconds of acquisition. BFV values were averaged across all pixels in each selected vessel segment and the mean BFV of a cardiac cycle was used for analysis. Pearson product moment correlation assessed the association between the two devices.
Results :
Significant association was found for BFV in the in-vitro setup between the two devices (Figure 2), r=0.80, p<0.001. In human subjects, the overall mean BFV across 63 vessels was 3.11±0.99 mm/s for the RFI and 9.35±1.28 a.u. for the XyCAM ID. The average CV was 30.5±10.6% for RFI measurements and 4.4±4.0% for the XyCAM ID measurements. The BFVs of the arterioles (3.14±0.91 mm/s, 9.36±0.93 a.u.) and venules (3.08±1.08 mm/s, 9.34±1.56 a.u.) were not significantly different, p>0.05 for the RFI and XyCAM ID respectively.
Conclusions :
Significant correlations were found between the two non-invasive retinal BFV measurement techniques for the in-vitro set up; however, due to the large CV, comparison of clinical data between the two devices was infeasible. The XyCAM ID, with a low CV of BFV measurements, may find application in research and eventually clinical diagnostics of retinal pathophysiology.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.