A Pixel Is Worth a Thousand Words

The Development of Ophthalmic Imaging and OCT

A Pixel Is Worth a Thousand Words

The Development of Ophthalmic Imaging and OCT.


Diagnostic imaging has undergone a dramatic change over the last 10 years. A multitude of new retinal therapeutics are changing clinical outcomes and offering new hope for disease states in which patients were historically destined to total loss of sight. The success and promise of a stream of new treatment discoveries are driving a parallel interest in new diagnostic imaging techniques.

New technologies such as spectral-domain optical coherence tomography (OCT) and autofluorescence are being used to see whether we can identify those patients most at risk, to measure specific response to therapeutics, and to look for ways to maximize potential combinations of therapy. The old saying "a picture is worth a thousand words" takes on new meaning as we apply what we see to how we see. Updating this adage for the recent innovations in digital imaging and scanning laser techniques, it may be more accurate today to say "a pixel is a worth a thousand words."

The transition from "picture" to "pixel" started at the University of Heidelberg, where Hermann von Helmholz developed the ophthalmoscope in 1851. With that, a new era in the understanding of the eye and retinal diagnostics began. In 1924, the emerging technology of photography merged with ophthalmoscopy based on the introduction of the Nordenson fundus camera, a portent of future successes that combine technologies. Matching cameras and film to record clinical observations enabled documentation of what the clinician saw, served as a baseline or reference for future change, and provided a tool for sharing the knowledge and experience with clinicians in training. As consumer awareness grew, the documentation served an important role for educating patients.

Jeffrey S. Heier, MD, practices ophthalmology with Ophthalmic Consultants of Boston and is president of the Center for Eye Research and Education Foundation in Boston. Dr. Heier reports the following moderate financial interest: Heidelberg Engineering (consulting/advisory fees and travel reimbursement)


It took time to transfer the benefits of photography to eye care, especially since one had to overcome the challenge of eye motion. Today, many people take flash photography for granted, both in consumer application and especially in ophthalmology. Although the flash provides illumination, helping to accurately record the look of the eye, it also critically stops eye motion. Most patients' eyes are in constant motion, stopping only for momentary fixation in milliseconds, then continuing to move in a constant search for new information.

Flash fundus photography helped create "freeze-frame" or stop-action photography, applying the principles of Eadweard J. Muybridge, who recorded the first images of a horse in gallop. In a well-known experiment, funded by philanthropist Leland Stanford, Muybridge used high-speed photography to capture the moment in time to prove that all 4 of a horse's hooves leave the ground during a gallop.

The combination of cameras and ophthalmoscopes continued as the mainstay of ophthalmic imaging for decades until a new form of light illumination appeared. The invention of the laser (an acronym for light amplification through the stimulated emission of radiation) in 1960 is credited to Theodore Maiman, but it was likely preceded by the work of Gordon Gould, who eventually won a long patent battle in 1986. Applying lasers to eye imaging helped overcome 2 very large problems: the first was the need to get a sufficiently large amount of light into the eye for recording a quality image, and the second was the need to rapidly capture an image in order to avoid blurring of the image due to eye motion.


Confocal scanning laser ophthalmoscopy (cSLO) was the first technical application of lasers for eye measurements. The first systems produced in commercial quantities were developed in Germany by Gerhard Zinser at Heidelberg Engineering in 1991 (earlier versions in prototype quantities were produced by Heidelberg Instruments and Rodenstock). The Heidelberg Retina Tomograph (HRT; Heidelberg Engineering, Vista, CA) used fast scanning speed to avoid motion artifact, capturing a complete transverse fundus image in 24 milliseconds. Involuntary microsaccades typically occur in 30 to 50 milliseconds, so the images were captured faster than the eye could move. The robustness of this strategy was born out in glaucoma by the clinical results demonstrated in the National Eye Institute-sponsored Ocular Hypertension Treatment Study. In that study, the HRT proved to be highly predictive of glaucoma up to 8 years in advance of detectable disease, based on visual fields and optic disc assessment.


Fluorescein angiography (FA) appeared at about the same time as the invention of the laser, with publication of the first article by Novotny and Alvis in 1960. However, it took many years for FA to become an accepted diagnostic indicator. The first images were believed to be interesting but were not well understood until Donald Gass published his landmark atlas.

After the appearance of the first cSLO, it was a natural step to adapt the cSLO to FA, with the first systems appearing in 1996. Using a laser emitting the peak excitation wavelength for fluorescein, the cSLO can produce images that appear different than photography or digital imaging but show greater detail. In addition, the high-speed lasers enable motion picture capture of the dye coursing through the vasculature, providing the clinician with an additional diagnostic element. This enables the clinician to better assess the vasculature surrounding the choroidal neovascularization, occasionally uncovering feeder vessels/retinal angiomatous proliferation.

This same strategy was later applied to indocynaine green angiography (ICGA), creating crisp clear images of the choroidal vasculature, thereby improving on the filtered flash photographs first used by Yannuzzi et al. Interestingly, ICGA and FA can be performed with a single injection and captured simultaneously on some imaging systems.


The role of cSLO technology was further expanded with the discovery of autofluorescence. Unlike FA or ICGA, no injection of dye is needed since cellular elements within the retinal pigment epithelium naturally fluoresce when exposed to a specific wavelength (488 nm) of light. Numerous studies have begun using autofluorescence images to track regions of geographic atrophy (GA) in dry age-related macular degeneration (AMD), testing the hypothesis that patterns of GA can be predictive of progression. Other studies are testing autofluorescence of GA as a surrogate endpoint for new therapeutics, such as the vitamin regimens being tested in Age-Related Eye Disease Study 2 trial.

Figure. An infrared image, taken with the Heidelberg Spectralis (left), and an OCT image, also taken with the Spectralis (right)


Originally developed by Fujimoto et al. at the Massachusetts Institute of Technology, the first commercial instrument using OCT was introduced by Carl Zeiss Meditec (Dublin, CA) in 1995. Instead of scanning transversely as with the cSLO technology, the OCT enabled a cross-sectional reconstruction of the retina using time-domain OCT, which could reveal the layers of the retina, often clearly showing the pathology (eg, the Figure).

The availability of this technology had a dramatic impact on retinal diagnosis and management of treatment alternatives. For example, OCT redefined the staging of macular holes, enabling differentiation of full thickness, partial thickness, pseudoholes, and cysts. In addition, the vitreoretinal interface of fellow eyes of these patients could be more easily assessed. For patients with edema, the OCT allowed rapid identification of fluid pockets, clearly showing if the fluid was intraretinal or subretinal.

The introduction of anti-vascular endothelial growth factor treatment increased the role and frequency of OCT imaging, enabling noninvasive monthly monitoring of agerelated macular degeneration patients. Recent data has suggested that OCT may be useful for determining treat/no-treat decisions.

Despite tremendous advances, time-domain OCT proved to have limitations, including a variety of artifacts ranging from eye-movement errors (due to slow scanning speeds) to operator errors based on software limitations that misidentified the fovea. Information regarding the nature and extent of artifact was slow to be recognized. The first articles commenting on OCT artifacts were published in engineering journalsm1,2 and only several years after product availability was the artifact problem presented in peerreviewed medical literature.3,4


The introduction of spectral domain OCT (SD-OCT) in 2006 promised to solve the eye movement problems of time-domain OCT and to give even higher resolution cross-sectional images by scanning at speeds 50 to 100 times faster. However, artifact can be seen even in 3-D reconstructions of the retina. Although scan times are significantly faster, full-volume scans (multiple stacks of B-scans) still need 1 to 3 seconds, which is not fast enough to avoid motion artifact. Still, the increase in image resolution and quality is significant and we are already beginning to redefine some of our impressions of disease with the new images.

A variety of strategies have been employed in an attempt to overcome the continuing problem of motion artifact. Most systems register a stack of B-scans against a fundus image taken either immediately before or after the OCT scan. Software algorithms then try to match the location of the B-scan and register it to the fundus image. One commercial system actually tracks the eye, using a dual-beam hardware configuration — one beam tracks the eye using a fundus image, then guides the second beam to place the OCT scan.

Another pressing issue is the utilization of OCT rather than FA as a monitoring and treatment tool in the evaluation of many of our injectable therapeutic advances. As retinal practices have become flooded with AMD patients needing monthly injections of ranibizumab (Lucentis, Genentech) or off-label bevacivumab (Avastin, Genentech), OCT has become the primary indicator of treatment due to the practical considerations of monthly monitoring in a noninvasive fashion.


With the abundance of new imaging modalities it is more likely that we will rely on a blend of indicators, rather than any single diagnostic, to help make the clinical decisions for treating our patients. There is a new world of diagnostics unfolding parallel to a host of new treatments making this one of the most exciting times in ophthalmology. RP

  1. Koozekanani D, Boyer KL, Roberts C. Tracking the optic nervehead in OCT video using dual Eigenspaces and an adaptive vascular distribution model. IEEE Trans Med Imaging. 2003;22:1519.
  2. Bartch D-UG, Gong X, Ly C, Freeman WR. Optical coherence tomography: interpretation artifacts and new algorithm. In: Fitzpatrick JM, Sonka M, eds. Medical Imaging 2004: Image Processing. Proceedings of SPIE. Vol. 5370. Bellingham, WA: SPIE; 2004:2140.2151.
  3. Ray R, Stinnett S, Jaffe GJ. Evaluation of image artifact produced by optical coherence tomography of retinal pathology. Am J Ophthalmol. 2005;139:18-2
  4. Sadda SR, Wu Z, Walsh AC, Richine L, Dougall J, Cortez R, LaBree LD. Errors in retinal thickness measurements obtained by optical coherence tomography. Ophthalmology. 2006;113:285-293.