Abstract
Purpose :
Bi-directional Doppler Optical Coherence Tomography (D-OCT) allows calculating absolute volumetric blood flow in retinal arterioles and veins. With two light beams, it renders flow results without knowing the angle between the absolute blood flow vector and the incident light beam, but it requires manual vessel segmentation. The aim of this study was to determine the reliability of bidirectional, swept source (D-OCT) retinal blood-flow measurements based on manual vessel segmentation and two different flow calculation methods.
Methods :
Doppler-OCT blood flow was measured using a prototype bi-directional swept-source 1050nm optical coherence tomography system. Four retina specialists were asked to independently delineate velocity signals on velocity-encoded B-scans. Based on these delineations, two different flow calculation methods were applied and compared: The first method is based on averaging the mean velocity per pixel and multiplying it by the pixel count of the region of interest, while the second one consists of the summation of all velocity values inside this region. The resulting flow values were compared along with their respective errors.
Results :
Four graders delineated the velocity signal of seven retinal arterioles and six retinal venules on two circular and three segment D-OCT scans. The mean delineated vessel diameter was 154.67 ± 57,51 microns (µm) with a mean relative error of 11.1 ± 3.42 %. The calculated average blood flow values of all delineated vessels were 12 ± 11.86 (1 ± 3.04 µl/min mean standard error, 17.3 ± 13.59 % relative error) with method 1 and 9 ± 7.21 µl/min (0 ± 0.21 µl/min standard error, 9.72 ± 7.21 % relative error) with method 2. In the subgroup of delineations sizes bigger than the mean, method 2 had a statistically significantly (p=0.008) lower (mean difference 8.378) relative error than method 1.
Conclusions :
Using bidirectional, dual-beam D-OCT data and manual signal delineation, a flow calculation based on summation includes less error while dealing with delineation differences than another method based on multiplication of the mean flow.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.