Human retinal oximetry using spectral imaging

• Main Author: Mordant, D J
• Format: Dissertation
• Language: English
• Published: ProQuest Dissertations & Theses 01-01-2012
• Subjects: Ophthalmology

The principal aim of the research described in this thesis was to develop a technique of non-invasively measuring the oxygen saturation within the retinal vasculature of human subjects (retinal oximetry). The evaluation of a hyperspectral fundus camera used to acquire retinal images in different wavelengths of visible light, and the image analysis techniques used to perform retinal oximetry are described. Validation of the oximetry techniques was performed using an artificial eye containing human blood of known oxygen saturation: the calculated oxygen saturation was compared to the gold standard measurement. The mean differences between the calculated and measured oxygen saturations were small. Hyperspectral imaging/oximetry of normal subjects was performed to characterize the oximetric features of the retinal vasculature. The mean oxygen saturation (± SD) of the temporal retinal arterioles and venules were 110.8% (± 11.8%) and 27.7% (± 3.2%) respectively. The application of the retinal oximetry technique was explored in patients with retinal arterial and venous occlusion to determine whether oximetric changes in the retinal vasculature could be detected. Variation in measured oxygen saturation of the retinal arterioles and venules respectively were apparent, and corresponded with angiographic features of retinal capillary loss. The techniques were applied to patients with asymmetrical primary open angle glaucoma to determine whether oximetric changes could be detected. The mean oxygen saturation of the temporal retinal venules were significantly higher [44.8% (± 24.2%)] in the more advanced glaucomatous eyes compared to normal subjects. Hyperoxia of the retinal venules suggests reduced oxygen consumption as a consequence of inner retinal dysfunction in glaucoma. However, because of the small sample size, further research on a larger population of subjects is required to support this finding. Hyperspectral imaging could be used to detect oximetric abnormalities in the retinal vasculature in patients with retinovascular occlusion and glaucoma.