Fundus Paintings and Photographs
Earliest forms of ocular imaging attempted to obtain representations of the ocular fundus as similar as possible to that seen by ophthalmoscopy. Paintings of various ocular diseases were used as plates in older texts, and often were renderings developed during interactions between an artist and an ophthalmologist. These paintings were time-consuming, expensive, idealized images. With the advent of fundus photography, specific images of patients could be recorded, stored, and indexed at low cost. Several important consequences arose:
- Individual patients could have more complete medical records.
- Interval changes in the fundus appearance for a specific patient could be accurately and objectively tracked.
- Later retrieval of the images could foster scientific research.
Simultaneous advances in camera design and film development lead to the possibility of monochromatic photography using special filters (often using green wavelengths) and color photography.
Ocular Angiography
Simple photography proved an efficient method of rendering topographic representations of the fundus, but not much else. The first ocular angiograms were made by a medical student, Harold Novotny, and an intern, David Alvis. They were trying to determine levels of oxygenation in the retina. Once they saw that with this technique they could document the circulation in the eye, they quickly began to investigate the new technique in people with diabetes and hypertension.
Fluorescein angiography opened dimensions in the study of retinal diseases that were previously unavailable. It became apparent that to fully understand the information available required integration of imaging information with knowledge about pathology and physiology. For example, breakdown of the blood ocular barrier allows leakage of fluorescein dye with pooling in cystoid spaces within the macula, making the picture of cystoid macular edema (CME). Although pattern recognition and the knowledge of pathology and physiology required to recognize CME is now well within the capabilities of a first year resident, the initial piecing together of all requisite retinal physiologic and pathologic information was not a simple matter. Once the connections were made, however, recognition became simple. To paraphrase Goethe, we see what we know.
Optical Coherence Tomography
In scientific investigation we often look at things we don't know. Because of the complexity and high visual cost of biopsy, ophthalmologists had difficulty appraising the structural changes in retinal diseases. Development of optical coherence tomography (OCT) enabled ophthalmologists to obtain cross-sectional images of the fundus, however, OCT didn't provide much information about the physiology of the pathologic process. To gain a more complete understanding of pathophysiologic processes, we had to integrate the information about structural changes observed with OCT with knowledge of physiology, pathology, ophthalmoscopy, fundus photography, ocular angiography, and in some instances, information gleaned from past experience with retinal surgery.
Fundus Autofluorescence
Fundus autofluorescence introduced a method to obtain not only topographic renderings of the fundus, much like what could be seen with ordinary fundus photography, but the brightness within the image originated from fluorophores in the eye that were produced and retained under a specific set of physiologic rules. Because the eventual accumulation of the fluorophores is integrally related to metabolism and physiology of the retina and retinal pigment epithelium, autofluorescence photography supplies functional information as well. Understanding autofluorescence photography required considering aspects of ocular physiology not previously encountered in clinical practice. This knowledge needed to be integrated into the conglomerate of knowledge required for proper interpretation of optical coherence tomography, ocular angiography, and fundus photography.
The Power and Complexity of Integrative Networks
Robert Metcalfe, one of the inventors of computer networking, proposed that the value of a network increased by the square of the number of compatibly communicating devices on that network. This idea has spread beyond computers to include other forms of interaction, including social networks. In this sense each unique source of information increases the value of the whole by more than only a linear increment; the increase in value is a power function.
In medicine knowledge is inevitably partial, and even that knowledge has been constructed as a part of a coherent network of mutually supporting beliefs and ideas. Our ability to evaluate, perceive, and think about ocular conditions and diseases is increased by the number of ways we can gain unique information. This is not merely obtaining more data; the information gained exists in an interlocking matrix. For every conclusion, diagnosis, or even unit of knowledge, each method of ocular imaging potentially contributes some amount. We could construct mathematical equations illustrating this by assigning information provided by each imaging method with a coefficient. There could be interactions among the variables as well. Each unique method of obtaining information supplies us with more variables, but also allows us to have more simultaneous equations to solve. This increase in complexity is daunting to some.
Early after the introduction of a new imaging modality, it is common for researchers to try to understand boundary conditions for the information the instrument provides. The simplest is to look for conditions in which the imaging modality used in isolation will provide practical information. For example, OCT can be evaluated as a test for CME. Although OCT would be compared with some gold standard test, such as fluorescein angiography or biomicroscopy, the goal would be to be able to use OCT in isolation to diagnose CME. In this circumstance we are looking for diseases for which we don't need to solve many simultaneous equations. If we had an equation relating the contributions from the imaging sources to the final required information to make a diagnosis, we would zero out the coefficients for the other modes of information except for that provided by OCT.
Another simple approach is to identify a particular disease in which an obvious incompleteness in knowledge exists because previous methods of examination provided no such information. In this situation we know that the coefficients for the older methods of imaging must be close to zero. Although we had ideas from biomicroscopy or fluorescein angiography, before the advent of OCT, we had incomplete and flawed ideas about macular hole formation. This is not to say that biomicroscopy or fluorescein angiography do not supply information about macular hole formation, only that the magnitude and significance of their coefficients are much less than those for OCT.
Another simple approach in imaging is to examine the outcomes in which we anticipate the results based on previous methods of examination. In this situation we have a working equation, and the coefficients for the other imaging modalities appear to have established nonzero values. If when using the new test we actually find the anticipated result, we may gain more confidence simultaneously in both our established theories and the validity of the new method of examination, reflecting the constructivist epistemologic nature of medicine. On the other hand finding significant deviations would provide both the reason and information to re-evaluate our currently held hypotheses. We would then re-evaluate both the coefficients and the results from the old equation. We anticipated that patients with branch retinal vein occlusion would have edema, but we didn't anticipate that a significant proportion would have a concurrent serous detachment of the macula when examined with OCT.
The more interesting, but ultimately more difficult, approach is to integrate information from more than one imaging modality to derive new information or new hypotheses. Newer investigations into diseases such as central serous chorioretinopathy or vitelliform macular dystrophy integrate data derived from fluorescein and indocyanine green angiography, OCT, and autofluorescence photography. Some ophthalmologists feel comfortable using one or a few imaging modalities because these tests define a circumscribed range of possibilities. The history of science is filled with examples of newer imaging techniques providing epiphanies, overcoming inadequacies of older methods. However each new method of imaging, by default, creates more questions than answers. Some people interested in imaging are willing to trade off the potential for new possibilities to avoid the uncomfortable feeling of uncertainty.
Author Disclosure
Dr. Spaide has received grant support from Genentech, Inc., consultant fees from Topcon Medical Systems, Glaxo Smith Kline, and Heidelberg Engineering, Inc., and royalties from DORC International and Topcon.