A regular SLO image can be distorted if eye movements occur during image acquisition. For the velocity measurements presented here the role of eye movements is minimized in several ways. First, the measurement time (related to the number of selected image lines) required for a single velocity estimate is about 5 ms on average, although up to 10 ms is required to measure the slowest cell velocities. Second, each frame includes an XY scanning portion, and therefore we can estimate how much the eye has moved during the 1/15 second between successive XY scans. For the results presented here we used data where the movements between frames were limited to a few microns (no saccades). Finally, during the XT portion of each frame we can detect motion of the eye based on the fact that stationary structures should appear as vertical straight lines. If there is a horizontal eye movement, then these lines will deviate from the vertical. If there is a vertical component of motion, and the stationary structures are small, then the scan line will be moved off the structure and the vertical lines will be interrupted. We restricted our data analysis to measurements taken where clean vertical straight lines were present, and interframe motions were minimal.
shows a movie of the blood flow within an artery of a healthy subject's retina, consisting of 40 frames acquired in 2.667 seconds. The artery is the vessel at point (a) in , whose lumen diameter is about 65 μm. The movie consists of XT images, along with an animation showing the velocity change. The blood velocity averaged over the vessel cross section is plotted over time, with the data points representing the calculated value from the consecutive XT images. The velocity fluctuates with the expected pulsatile pattern, rising rapidly at the beginning of a cardiac cycle and dropping slowly afterward. The velocity waveform shows a dicrotic notch after the systolic peak, which is very typical in arterial pressure waveforms and is thought to result from the closure of the aortic valve[23
]. The fluctuation repeats 3 times in the 2.667 seconds, matching closely the subject's heartbeat period which was measured immediately after data acquisition.
(Media 2) Movie showing fluctuating blood velocity during the subject's cardiac cycle. The blood vessel is a medium-sized artery.
Locations of arterial blood flow measurements before (a) and after a bifurcation (b & c). Scale bar is 100 μm.
shows data from a retinal vessel, where the scanning line crosses the vessel axis at a very small angle. The left part of the XT image is taken at a region near the vessel wall, while the right part is at the center of the vessel. The corresponding calculated velocity profile is shown as a bar plot. The steeper slopes on the left show that cells near the vessel wall are moving more slowly than cells near the central lumen, as expected. The velocity profile can only be measured in blood vessels which cross the scanning line with a small angle, which depends on the blood vessel orientation and is not controlled in the current system. Although the cell velocity profile shown here is close to the left side of a blunted parabolic profile, we do not have sufficient data to test the flow model used to calculate flow in this paper.
Fig. 6 Change of erythrocyte velocity across the vessel lumen. The different slope angles in the XT image represent different velocities, with steeper slopes indicating lower velocities. At the left and right side of the XT image, the scan line is near the vessel (more ...)
To test the consistency of the measurement of blood flow using our method, we measured the total flow in the branches of an artery before and after a bifurcation, at least 5 vessel diameters away from the bifurcation to avoid non-laminar blood flow conditions. The results are displayed in . shows the location of the blood vessels. The flow rate was measured consecutively, first in the parent vessel and then in the daughter vessels. Each measurement took 5 minutes during which we assume that the actual average blood flow rate remained the same. The calculated time-averaged blood flow rate before the bifurcation at position (a) was 2.30 μl/min and the sum in the two daughter vessels at positions (b) and (c) after the bifurcation was 2.37 μl/min.
Calculated results of the vessel branches in
The system as currently implemented does have some limitations. First, our velocity measurements are required to be taken in the plane of the blood vessel. If a blood vessel were oriented at an angle in depth, we would underestimate the velocity. This is not a major problem, because most of the vessels in the size range of interest lie approximately in the plane of retina, with connections between vascular layers being primarily vertical and readily visible since they go out of focus[24
]. The second limitation relates to the size of vessels we can measure. We have not been able to reliably measure flow in the smallest capillaries, due in part to the low contrast. Similarly for the largest retinal vessels, very high blood velocities are not readily measured due to frequency limitations of the horizontal scanner in our system (8 kHz). This scan frequency represents a fundamental sampling rate for this technique, just as it does for the similar approach used by the Heidelberg flowmeter[25
]. As a result, this present system cannot accurately measure the highest blood velocities in blood vessels with diameters larger than 100 μm.
There are two approaches to addressing the limitation to the upper end of the velocity range of this technique. The first is simply to use a faster horizontal scanner. The second approach is to control the angle at which the scanning line crosses the blood vessel. The main limitation is that a flowing erythrocyte must take sufficient time to cross the scanning line so that there are enough samples to estimate the slope of the streaks. The time to cross the scanning line is inversely proportional to the component of the erythrocyte velocity perpendicular to the fast scanning direction. Thus, decreasing the angle α between the scanning line and the vessel axis will decrease this velocity component and increase the time during which a given erythrocyte is visible in the XT image and allow higher velocities to be measured. This also has the advantage that it would allow better measurements of cell velocity and the related velocity profile on more blood vessels with a more complete range of orientation angles, since the third limitation of the current system is that it cannot take accurate measurements in blood vessels which cross the scanning line at a large angle. This would require rotating the scan direction with respect to the blood vessel orientation to select an optimal angle for a given vessel, which is technically feasible using an image rotator or rotation of the galvanometer.