|
We have developed a snapshot optical imaging system capable of taking multiple images simultaneously and
sending them to a CCD detector. Using an innovative lens array design, the beam obtained from the fundus camera is
segmented into several different images passed through several different bandpass filters. Each bandpass filter defines a
unique spectral region of imaging. The images are taken simultaneously into a large silicone chip with a dynamic range
of 16 bits (highly sensitive) and are integrated with a single optical connection to a digital fundus camera.
Our algorithm maps blood oxygen saturation of the retina using several wavelengths. These wavelengths are
capable of approximating the whole hemoglobin spectrum and have been found from a previously developed
hyperspectral algorithm. They include four isosbestic points (522, 548, 569, and 586 nm) and three oxygen-sensitive
points (542, 560, and 586 nm) where the difference between fully oxygenated and deoxygenated blood is at a maximum.
Using MatLab code, color maps of oxygen saturation are produced. The average value taken from all vein areas was 60.53%, assuming that the artery oxygen saturation value is
98%. Oxygen saturation of the tissue was 75.78%. Oxygen saturations of the temporal/inferior/nasal veins ranged from
61.86% to 63.37%; the superior vein was significantly lower (54.19%). Tissue oxygen saturations in different quadrants
of the eye ranged from 74.17% to 76.74%.
Our algorithm has been developed for measuring oxygen saturation of the retina clinically. This was done for
one subject only; further work can extend the measurements to different pigments.
|
