Four-wavelength retinal vessel oximetry

This dissertation documents the design and construction of a four-wavelength retinal vessel oximeter, the Eye Oximeter (EOX). The EOX scans low-powered laser beams (at 629, 678, 821 and 899 nm) into the eye and across a targeted retinal vessel to measure the transmittance of the blood within the vessel. From the transmittance measurements, the oxygen saturation of the blood within the vessel is computed. Retinal vessel oxygen saturation has been suggested as a useful parameter for monitoring a wide range of conditions including occult blood loss and a variety of ophthalmic diseases.;An artificial eye that simulates the geometry of a human retinal vessel was constructed and used to calibrate the EOX saturation measurement. A number of different oximetry equations were developed and tested. From measurements made on whole human blood in the artificial eye, an oximetry equation that places a linear wavelength dependance on the scattering losses (3% decrease from 629 to 899 nm) is found to best calibrate the EOX oxygen saturation measurement. This calibration also requires that an adjustment be made to the absorption coefficient of hemoglobin at 629 nm that has been reported in the literature.;More than 4,000 measurements were made in the eyes of three human subjects during the development of the EOX. Applying the oximetry equation developed through the in vitro experiments to human data, the average human retinal venous oxygen saturation is estimated to be 0.63 +/- 0.07 and the average human retinal arterial oxygen saturation is 0.99 +/- 0.03. Furthermore, measurements made away from the optic disk resulted in a larger variance in the calculated saturation when compared to measurements made on the optic disk.;A series of EOX experiments using anesthetized swine helped to verify the sensitivity of the EOX measurement of oxygen saturation. It is found that the calibration in swine differed from the calibration in the artificial eye. An empirical calibration from the swine measurements was applied to the human measurements. With this correction, the average human retinal venous oxygen saturation was found to be 0.40 +/- 0.08 and the average human retinal arterial oxygen saturation was found to be 0.98 +/- 0.03. This suggests that the swine, human and model eye measurements may each require a different calibration because of differences between the spectral characteristics of their ocular fundi. Finally, further in vitro and in vivo experimentation is proposed so that the goal of having an accurate, absolute measurement of the human retinal vessel oxygen saturation will be realized.