Abstract
Purpose: :
To present the latest progress in quantitative retinal blood flow measurements, system calibration, and automatic retinal blood vessel parameter extraction with ultrahigh resolution (~3 µm) spectral domain OCT (SD-OCT).
Methods: :
We have developed algorithms for measuring the retinal blood flow with SD-OCT, which requires the calculation of the absolute blood flow velocity and vessel diameters. We designed a specific arc shaped scan pattern for imaging individual retinal blood vessels including a scan for eye motion compensation. To calculate the absolute flow velocity the orientation of the vessel need to be extracted from the OCT images, which includes the extraction of the lateral and depth coordinates of the vessels. The lateral coordinates were calculated by using the concept of OCT fundus shadowgram, a technique for generating high contrast fundus image first developed by our group, with a series of signal processing. After determining the lateral coordinates and the vessel diameter the depth coordinate of the vessels in each OCT image was calculated from the corresponding Doppler image generated for each of the arc scans. To verify the accuracy of this method a phantom simulating a blood vessel with adjustable 3D orientation and flow rate was built and tested.
Results: :
The method was tested on selected retinal vessels of normal human eyes non-invasively. A study on a retinal artery yielded a Doppler angle of 87.8º, a blood flow velocity of 30.4±9.5 mm/s (mean and standard deviation) and an average blood flow of 17 µl/min. The results of measurements for a retinal vein revealed a Doppler angle of 87º, a blood flow velocity of 16.4±3.9 mm/s and an average blood flow of 2.01 µl/min. These results compared well with previously published data. The tests with the blood vessel phantom showed that the technique has an error of less than 10% compared with the actual set flow.
Keywords: image processing • vascular occlusion/vascular occlusive disease • blood supply