The purpose of this study was to measure the hemoglobin oxygenation in retinal vessels and to evaluate the sensitivity and reproducibility of the measurement. Using a fundus camera equipped with a special dual wavelength transmission filter and a color charge-coupled device camera, two monochromatic fundus images at 548 and 610 nm were recorded simultaneously. The optical densities of retinal vessels for both wavelengths and their ratio, which is known to be proportional to the oxygen saturation, were calculated. From 50-deg images, the used semiautomatic vessel recognition and tracking algorithm recognized and measured vessels of 100 microm or more in diameter. On average, arterial and venous oxygen saturations were measured at 98+/-10.1% and 65+/-11.7%, respectively. For measurements in the same vessel segments from the five images per subject, standard deviations of 2.52% and 3.25% oxygen saturation were found in arteries and veins, respectively. Respiration of 100% oxygen increased the mean arterial and venous oxygen saturation by 2% and 7% respectively. A simple system for noninvasive optical oximetry, consisting of a special filter in a fundus camera and software, was introduced. It is able to measure the oxygen saturation in retinal branch vessels with reproducibility and sensitivity suitable for clinical investigations.

Functional alterations are first signs of a starting pathological process. A device that measures parameter for the characterization of the metabolism at the human eye-ground would be a helpful tool for early diagnostics in stages when alterations are yet reversible. Measurements of blood flow and of oxygen saturation are necessary but not sufficient. The new technique of auto-fluorescence lifetime measurement (FLIM) opens in combination with selected excitation and emission ranges the possibility for metabolic mapping. FLIM not only adds an additional discrimination parameter to distinguish different fluorophores but also resolves different quenching states of the same fluorophore. Because of its high sensitivity and high temporal resolution, its capability to resolve multi-exponential decay functions, and its easy combination with laser scanner ophthalmoscopy, multi-dimensional time-correlated single photon counting was used for fundus imaging. An optimized set up for in vivo lifetime measurements at the human eye-ground will be explained. In this, the fundus fluorescence is excited at 446 or 468 nm and the time-resolved autofluorescence is detected in two spectral ranges between 510 and 560 nm as well as between 560 and 700 nm simultaneously. Exciting the fundus at 446 nm, several fluorescence maxima of lifetime t1 were detected between 100 and 220 ps in lifetime histograms of 40 degrees fundus images. In contrast, excitation at 468 nm results in a single maximum of lifetime t1 = 190 +/- 16 ps. Several fundus layers contribute to the fluorescence intensity in the short-wave emission range 510-560 nm. In contrast, the fluorescence intensity in the long-wave emission range between 560 and 700 nm is dominated by the fluorescence of lipofuscin in the retinal pigment epithelium. Comparing the lateral distribution of parameters of a tri-exponential model function in lifetime images of the fundus with the layered anatomical fundus structure, the shortest component (t1 = 190 ps) originates from the retinal pigment epithelium and the second lifetime (t2 = 1,000 ps) from the neural retina. The lifetime t3 approximately 5.5 ns might be influenced by the long decay of the fluorescence in the crystalline lens. In vitro analysis of the spectral properties of expected fluorophores under the condition of the living eye lightens the interpretation of in vivo measurements. Taking into account the transmission of the ocular media, the excitation of NADH is unlikely at the fundus.

The increase of retinal vessel oxygen saturation in diabetic retinopathy points to a diabetic microvascular alteration. This may be due to occlusions and obliterations in the capillary bead and the formation of arterio-venous shunt vessels. On the other hand, hyperglycaemia-induced endothelial dysfunction, with subsequent suppression of the endothelial NO-synthase and disturbance of the vascular auto-regulation, may contribute to retinal tissue hypoxia.

Various models have been published calculating the light transport at the ocular fundus either for interpretation of in vivo reflectance measurements or for the prediction of photocoagulation effects. All these models took the absorption spectra of the pigments located at the ocular fundus, melanin, haemoglobin, xanthophyll, and the photoreceptor pigments, into account. However, light scattering inside the single fundus layers has not been investigated in detail and was, therefore, neglected in the calculations or only considered by very rough approximations. This paper presents measurements on specimens of retina, retinal pigment epithelium, choroid, and sclera using the double-integrating-sphere technique. Absorption coefficients, scattering coefficients, and anisotropy of scattering were calculated by an inverse Monte Carlo simulation from the measured collimated and diffuse transmittance and diffuse reflectance. Conclusions are drawn for the interpretation of fundus reflectance measurements, which are a useful tool in diagnostics and photocoagulation dosimetry.

A new method for the spatially resolved measurement of the oxygen saturation of retinal vessels is described. Imaging spectrometry was used for both measurements of transmission and reflectance spectra of whole blood in cuvettes as well as for fundus reflectance spectra. A model was developed for the calculation of the oxygen saturation, valid in the wavelength range between 510 nm and 586 nm, in that the internal reflectance is constant and only the transmitted light depends on layer thickness and hematocrit. Altogether 265 measurements were performed in different number at 30 eyes. In each measurement, the oxygen saturation was simultaneously determined for 193 locations along a line of 1.5 mm at the fundus. The mean oxygen saturation in retinal arteries was (92.2 +/- 4.1)% and (57.9 +/- 9.9)% in retinal veins. The mean retinal arterio-venous difference of the oxygen saturation was (35.1 +/- 9.5)%. The venous oxygen saturation depended on distance from the optic disc. The measured mean of the arterio-venous difference of the oxygen saturation corresponded well to the value of the brain (34%). The utilization of oxygen in the temporal quadrants (inferior: 39.4 +/- 10.4%) is significantly (p = 0.05) higher than in the nasal quadrants (inferior: 31.3 +/- 6.7%).