Abstract
Purpose: :
Spectral imaging of the retina enables non-invasive oximetry of blood within retinal vessels. This is particularly relevant to research and clinical detection of retinal diseases.We report oxygen saturation mapping of the retina by means of a clinical tool comprising of a unique snapshot spectral retinal imager and novel algorithms for quantitative analysis.
Methods: :
An Image Replicating Imaging Spectrometer (IRIS) device has been fitted in a commercial fundus camera which enables simultaneously recording of eight wide-field spectral images of the retina onto a single detector. The snapshot capability removes the fundamental problems inherent in time-sequential techniques which affect accurate blood oximetry. The spectral pass-bands for IRIS have been optimised for blood oximetry and are incorporated into the physical model of light propagation in the retina. Algorithms for semi-quantitative and quantitative spectral analysis have been developed which employ some or all of the eight spectral images.
Results: :
Semi-quantitative and quantitative oximetric maps have been calculated by spectral unmixing and physical-model-based techniques respectively, as shown in the figure below (oxygenated arterial blood is false-coloured red, whilst deoxygenated venous blood is coloured black). Results obtained with both methods are in good agreement. An artificial retina containing blood of known oxygenation has been used to validate these techniques.
Conclusions: :
A snapshot spectral imaging technique has been developed together with spectral algorithms for accurate oximetry and usability in a clinical instrument. This technique shows great promise for the early detection of retinal disease in retinal screening programs.
Keywords: retina • blood supply • oxygen