Automatic Retinal Oximetry Sveinn Hakon Hardarson,1 Alon Harris,2 Robert Arnar Karlsson,3 Gisli Hreinn Halldorsson,3 Larry Kagemann,2,4 Ehud Rechtman,2 Gunnar Ma´r Zoega,1 Thor Eysteinsson,1 Jon Atli Benediktsson,3 Adalbjorn Thorsteinsson,5 Peter Koch Jensen,6 James Beach,7 and Einar Stefa´nsson1 PURPOSE. To measure hemoglobin oxygen saturation (SO2)in noninvasive device to measure retinal oxygen levels is retinal vessels and to test the reproducibility and sensitivity of Aneeded for the study of the role of oxygen levels in human an automatic spectrophotometric oximeter. retinopathy and glaucoma. Spectrophotometric measurements METHODS. Specialized software automatically identifies the ret- of hemoglobin oxygen saturation in human retinal blood ves- 1 inal blood vessels on fundus images, which are obtained with sels was initiated by Hickam et al. in the 1960s but has still not four different wavelengths of light. The software calculates been mastered as a reliable and useful clinical or research tool. optical density ratios (ODRs) for each vessel. The reproducibil- During the past 7 years, we have built on the work of Beach et ity was evaluated by analyzing five repeated measurements of al.,2 Delori,3 and others4 to develop a noninvasive automatic the same vessels. A linear relationship between SO2 and ODR retinal oximeter, which we hope will allow studies of human was assumed and a linear model derived. After calibration, retinal and optic nerve disease in research and clinical settings. reproducibility and sensitivity were calculated in terms of SO2. We report the sensitivity and reproducibility of this device in Systemic hyperoxia (n ϭ 16) was induced in healthy volun- human studies. 5 teers by changing the O2 concentration in inhaled air from 21% Michaelson (1948) suggested that hypoxia plays a role in to 100%. the pathophysiology of diabetic and other retinopathies. Ani- RESULTS. The automatic software enhanced reproducibility, and mal research and limited studies on humans suggest that oxy- the mean SD for repeated measurements was 3.7% for arte- gen may be an important factor in the pathogenesis and treat- 6–11 rioles and 5.3% venules, in terms of percentage of SO2 (five ment of some retinal diseases. These include venular repeats, 10 individuals). The model derived for calibration was (Scibor M et al. IOVS 2002;43:ARVO E-Abstract 3305) and ϭ Ϫ ⅐ Ϯ 12,13 14–19 SO2 125 142 ODR. The arterial SO2 measured 96% 9% arteriolar occlusions, retinal detachment, and dia- (mean Ϯ SD) during normoxia and 101% Ϯ 8% during hyper- betic retinopathy20–35 (Trick GL et al. IOVS 2005;46:ARVO oxia (n ϭ 16). The difference between normoxia and hyper- E-abstract 1413). Animal studies have also indicated that both oxia was significant (P ϭ 0.0027, paired t-test). Corresponding vitrectomy7,36–38 and laser photocoagulation8,9,28,36,38–45 numbers for venules were 55% Ϯ 14% and 78% Ϯ 15% (P Ͻ raise the oxygen tension in the retina. 0.0001). SO2 is displayed as a pseudocolor map drawn on Although glaucoma therapy is currently aimed at lowering fundus images. intraocular pressure, the beneficial effects may be due to better CONCLUSIONS. The retinal oximeter is reliable, easy to use, and perfusion and oxygenation of the optic nerve rather than the sensitive to changes in SO2 when concentration of O2 in direct effects of lower intraocular pressure. Experiments in inhaled air is changed. (Invest Ophthalmol Vis Sci. 2006;47: monkeys,46,47 cats,48,49 rabbits,50 and pigs51,52 have demon- 5011–5016) DOI:10.1167/iovs.06-0039 strated that optic nerve oxygenation decreases when the per- fusion pressure declines and autoregulation of blood flow is overwhelmed (e.g., because of raised intraocular pressure). From the Departments of 1Ophthalmology and 5Anesthesiology, Poor regulation of blood flow has been postulated as the 53,54 University of Iceland, National University Hospital, Reykjavı´k, Iceland; mechanism of glaucomatous optic atrophy. the 2Department of Ophthalmology, Indiana University/Purdue Univer- These theories are predominantly based on animal research, sity School of Medicine, Indianapolis, Indiana; the 3Faculty of Electrical and few studies have been conducted to investigate oxygen- 4 and Computer Engineering, University of Iceland, Reykjavik; the De- ation in humans, because reliable and noninvasive methods of partment of Ophthalmology, University of Pittsburgh School of Medi- measuring human retinal oxygenation have been lacking. In cine, Pittsburgh, Pennsylvania; the 6Eye Department, National Univer- 7 the past decades, various approaches to retinal oximetry have sity Hospital of Copenhagen, Denmark; and the Institute for 4 Technology Development, Stennis Space Center, Mississippi. been tried (for a review of the methods, see Harris et al. ). Most Supported by Icelandic Research Council Grants 050409011 and of these methods are based on the fact that oxygenated and 051220005, The Landspı´tali—University Hospital Fund for Science, deoxygenated hemoglobin have different light absorption The Research Fund of The University of Iceland, and The Icelandic spectra (color). By analyzing the light absorbance of blood at Fund for Prevention of Blindness. two or more wavelengths, the oxygenation of hemoglobin can Submitted for publication January 13, 2006; revised May 9 and July be estimated. Investigators in this field have run into numerous 19, 2006; accepted September 18, 2006. Disclosure: S.H. Hardarson, Oxymap (E, P); A. Harris, None; problems, however, that have prevented the development of a R.A. Karlsson, Oxymap (E, P); G.H. Halldorsson, Oxymap (E, P); L. reliable and practical method. The problems include nonlinear Kagemann, None; E. Rechtman, None; G.M. Zoega, None; T. Ey- sensors (photographic film), eye movement during lengthy steinsson, Oxymap (I, P); J.A. Benediktsson, Oxymap (I, P); A. measurements, small areas measured at one time, and optical Thorsteinsson, None; P.K. Jensen, None; J. Beach, Oxymap (I, P); complexities within the eye. Nevertheless, considerable E. Stefa´nsson, Oxymap (I, P) progress has been made.2,3,55,56 The publication costs of this article were defrayed in part by page In this study we compared the reproducibility of our man- charge payment. This article must therefore be marked “advertise- ment” in accordance with 18 U.S.C. §1734 solely to indicate this fact. ual and automatic methods for retinal oximetry. We present Corresponding author: Einar Stefa´nsson, Department of Ophthal- calibration for our automatic method and use this to display the mology, University of Iceland, National University Hospital, Reykjavı´k, reproducibility and sensitivity of our retinal oximeter to hemo- Iceland; [email protected]. globin oxygen saturation. Investigative Ophthalmology & Visual Science, November 2006, Vol. 47, No. 11 Copyright © Association for Research in Vision and Ophthalmology 5011 Downloaded from iovs.arvojournals.org on 09/25/2021 5012 Hardarson et al. IOVS, November 2006, Vol. 47, No. 11 METHODS In equation 2, SO2 is the percentage of hemoglobin oxygen saturation; a and k are constants; ODX and ODY are ODs (no unit) at wavelengths All procedures were approved by the institutional review boards of the X and Y, respectively; and ODR is the optical density ratio. Thus, in respective institutions: Indiana University-Purdue University, Indianap- theory, hemoglobin oxygenation can be calculated using brightness olis, and the University of Iceland, National University Hospital, Reyk- inside and outside vessels at two wavelengths of light. javı´k. Informed consent was provided by all subjects before participa- tion in the studies. All procedures conformed to the tenets of the The Software Declaration of Helsinki. Two versions of software were used. Both versions register all four output images in the same coordinate system, so that each pair of The Retinal Oximeter points, used for calculation of ODs, is at the same fundus location for all four images. The registration algorithm utilizes the edges of the four Our method for retinal oximetry is based on the work of Beach et al.2 images.57 The older software version requires the user to select mea- The setup and outcome are shown in Figure 1. surement points manually, inside and outside vessels to calculate the Two retinal oximeters were used for the study, one based in ODR. The newer version automatically locates vessels on the fundus Indianapolis and the other in Reykjavik. Both are composed of a fundus image and selects measurement points inside and outside vessels. camera coupled with a beam splitter to a digital camera. (In Indianap- Morphologic filters are used to enhance vessellike structures in the olis, the fundus camera was a Topcon TRC50-VT; Topcon Co., Tokyo, images. After morphologic enhancement, a skeleton of a thresholded Japan; the beam splitter was a MultiSpec Patho-Imager; Optical In- fundus image is used to locate the approximate centerlines of the sights, Tucson, AZ; and the digital camera was a CoolSNAP ES; Roper arterioles and venules. To avoid error caused by specular reflection, Scientific, Tucson, AZ. In Reykjavik, the same beam splitter was used, the new software uses the darkest point on the vessel’s cross section. but the fundus camera was a Canon CR6-45NM; Canon Inc., Tokyo, To reduce variation in calculated ODRs due to random system noise Japan; the digital camera was an SBIG ST-7E; Santa Barbara Instrument and variable optical properties of the fundus, the software averages Group, Santa Barbara, CA, used at 2 ϫ 2 binning). The beam splitter measurements from several pixels adjacent to the vessel. When the separates the original image into four optical channels. In each chan- user has specified the vessel segment of interest, the mean ODR for nel, there is a different narrow band-pass filter, through which only that segment is given. The automatic software also displays a color- light of specific wavelengths can pass. The center wavelengths of the coded map of the vessels, where the colors indicate hemoglobin filters are 542, 558, 586, and 605 nm and the half-bandwidth is 5 nm, oxygen saturation (after calibration, described later).