Fundamentals of Fundus Autofluorescence Imaging

New imaging technology reveals a biomarker of retinal disease progression that's not visible to the clinician's eye.

By Khadija Shahid, OD

Release Date: January 2013
Expiration Date: January 1, 2016

Goal Statement:

The evolution of ophthalmic imaging has coincided with, and aided, the management of advanced ocular disease. Fundus autofluorescence (FAF) imaging is a relatively new technology that, along with other diagnostic tools, can help the clinician acheive a better understanding of the health of the fundus in order to obtain earlier diagnoses and better predict progression of certain retinal diseases. This course explains how FAF works, and how its results are interpreted, for a variety of retinal disease conditions.

Faculty/Editorial Board:

Khadija Shahid, OD

Credit Statement:

This course is COPE approved for 2 hours of CE credit. COPE ID 36505-PS. Check with your local state licensing board to see if this counts toward your CE requirements for relicensure.

Joint-Sponsorship Statement:

This continuing education course is joint-sponsored by the Pennsylvania College of Optometry.

Disclosure Statement:

Dr. Shahid has no relationships to disclose

Lipofuscin (LF) is a byproduct of phagocytosed photoreceptor outer segments that accumulates in the retinal pigment epithelium (RPE) with age as well as in certain retinal diseases. When exposed to short- to medium-wavelength visible light, LF will autofluoresce. Fundus autofluorescence (FAF) imaging takes advantage of the autofluorescent properties of LF to document its accumulation. The FAF imagery can then be used to predict patterns of disease and progression and can lead to better understanding of disease pathogenesis.

By bridging FAF with additional imaging technologies—including digital red, green, blue (RGB) monochromatic filters, topographical emboss filters and image registration technology, along with high-resolution optical coherence tomography (OCT)—optometrists can further evaluate the ocular fundus in detail to:

  • Track temporal changes in LF distribution.
  • Detect earlier certain retinal disorders related to LF accumulation.
  • Assess risk factors that may affect LF accumulation in the fundus.
  • Aid in differentiating diseases using specific LF accumulation patterns.

The use of these advanced imaging systems offer primary care optometrists further understanding of retinal disease pathogenesis, and may aid in counseling of preventative steps for enhanced patient management.

Imaging History

Retinal photography has evolved rapidly since its inception in 1959 using an electronic flash tube and 35mm black-and-white film-based fluorescein angiography. By the 1980s, digital retinal imaging became available using charged couple device (CCD) light sensors. More sensitive CCDs, as well as wavelength extraction and digital processing software, have created a revolution in ocular digital imaging.

The 1990s saw the introduction of OCT technology. As high-resolution scanning laser ophthalmoscope (SLO) OCT became more accessible, so did optimal, advanced management of ocular disease.

In the past decade, a new addition to the imaging front is the use of fundus autofluorescence technology.

What is Lipofuscin?

Lipofuscin is a biomarker evident in normal aging and in chronic disease. Its accumulation has been detected in various tissue and lesions associated with neurodegeneration (Parkinson's, Alzheimer's, etc.), nutritional cirrhosis, cardiac failure and RPE degeneration underlying retinal disease, among many other conditions.1

This accumulation is evident in ocular disease even before the visual cycle begins to degrade, supporting the theory that LF accumulation can be viewed as an early marker of certain retinal degenerative processes.1,2

How Do We Image Lipofuscin?

RPE lipofuscin deposits are visual pathway byproducts with unique autofluorescent properties that can be detected and quantified using imaging devices such as the fundus spectrophotometer, confocal scanning laser ophthalmoscope (cSLO) and FAF camera systems—with the latter two being the most commonly used clinical instruments.3 Because the approximate spectrum range of LF (wavelength range: 300nm to 600nm) is close to the visible spectrum of light (wavelength range: 400nm to 700nm), these clinical instruments can use visible light to elicit an emission, and safely detect LF in vivo during a routine clinical examination.1,3

* cSLO. The cSLO systems elicit and capture the RPE LF response by using a low-energy laser to excite LF, and a barrier filter to allow only the RPE LF response to pass. The cSLO can perform as many as 30 scans, which are then averaged together using post-processing software.4 The final result is a single, high-contrast, monochromatic image. The confocal optics and scanning laser help to bypass most of the anterior autofluorescence—for example, in an aging lens—which could interfere with posterior pole imaging.

* FAF camera. Alternately, a FAF retinal camera system uses a high-energy white flash (300 watt-seconds) and a wideband exciter filter to penetrate ocular media, reach deep within the RPE, and excite any existing LF. The LF response is then able to pass through a wideband barrier filter before reaching the sensor of the retinal camera. Similar to cSLO systems, the result is a single monochromatic image that reveals either the presence or absence of LF. The difference, however, is that FAF images from the retinal camera are not averaged—rather, a single image is captured in real time.

It's important to note that FAF results from both systems have quality limiting factors. As mentioned previously, ocular media opacities, such as an aging lens, can alter or negatively affect FAF posterior pole image results. Additionally, there is neither a uniform protocol (correction of patient refractive error, vertex distance, etc.), nor a standardized manufacture setting that dictates excitation and barrier filter wavelength setting or image processing techniques.

While all FAF systems require digital processing to create the final result, it's important to remember that FAF imaging provides quick, non-invasive access to information related to the health of RPE cells in relationship to LF.

Interpreting FAF results

FAF imagery used to detect and track changes in RPE LF must be interpreted appropriately to best understand ocular health status and to convey this to our patients.

Visually, FAF imagery resembles a fluorescein angiogram (FA) study in that results are represented by a 256 grey-scale value. Low pixel values represent low fluorescent intensities and appear dark, or hypofluorescent. Alternately, high pixel values appear bright, or hyperfluorescent. Unlike FA studies, where signal intensity is determined by circulation, the FAF signal is dependent solely on the presence of autofluorescent material (i.e., LF). Increased concentrations of LF result in very bright, hyperfluorescent signals. Conversely, in the absence of LF, signals appear dark (hypofluores-cent).

FAF imaging has been used in healthy subjects as young as two years old. The posterior pole of a healthy ocular fundus has an overall diffuse, mildly hyperfluorescent signal due to the normal levels of LF present in RPE cells. The optic nerve head always appears dark (hypofluorescent) due to the absence of RPE and LF (figure 1). Other structures that appear dark are retinal blood vessels (due to signal absorption from blood) and the fovea (due to signal absorption from high densities of macular luteal pigment).



The posterior pole of an unhealthy ocular fundus will have areas of abnormal signal densities (figures 2a,b). This could include hypofluorescent signals such as those seen with RPE atrophy and cell death, fresh hemorrhages, exudative lesions, areas of dense hyperpigmentation, and some forms of hard drusen. It also could include hyperfluorescent signals, such as those seen with abnormally high concentrations of RPE LF; for example, visible yellow lesions associated with lipofuscinopathy diseases (Best's, Stargardt's, etc.) are often intensely bright on FAF imaging due to abnormally high levels of LF. Examples of a more mild hyperfluorescent FAF signal could include older hemorrhages (due to fluorphore buildup within the stagnant blood), large, confluent, soft drusen, and basal laminar or reticular drusen that have a unique fluorescent pattern.



When comparing FAF patterns to the corresponding color image patterns of ocular disease, there can be large variability in the findings. The Fundus Autofluorescence in Age-related Macular Degeneration (FAM) Study described these variables in patients during an international workshop on FAF phenotyping for early AMD. Subjects were classified into one of eight different phenotypes based on different patterns of autofluorescence.5 (See "FAF Phenotypes in Early AMD".) The classification scheme illustrates the wide diversity of FAF patterns that are present in just one single disease: early AMD.



Among the important conclusions of this study: Visible alterations seen on color fundus photography were often poorly correlated with FAF imaging, and these FAF pattern differences were likely indicative of disease progression not yet visible to the clinician's eye. (See "FAF and Color Fundus Images in Ocular Pathology") In other words, FAF findings could represent an independent measure of disease activity.6



Let's look at the use of FAF for different disease conditions.

* AMD. Comparisons of FAF patterns and color imagery in late-stage, dry AMD with geographic atrophy (GA) also demonstrate a variety of FAF phenotypes that are not evident on color fundus photography or other imaging methods (figures 3a,b).7 A classification scale developed by the FAM Study group (See "FAF Phenotypes in Late AMD,") relates to FAF patterns at the junctional zone—the area that encompasses the border between the unaffected retina and the edge of a GA lesion. Those cases where more active, larger and more diffuse hyperfluorescent lesions were noted at the junctional zone were more likely to progress over time than those with absent or hypo-fluorescent lesions at the junctional zone. The progression of GA lesions seemed to be more dependent on the FAF pattern at the junctional zone than any other risk factor being monitored—including size of baseline atrophy, history of smoking, hypertension, diabetes, age, hyperlipidemia and family history.7 This may help our understanding of unpredictable prognoses in patients with AMD who present with similar baseline clinical findings, yet progress at quite dissimilar rates over the course of their disease. FAF imaging is likely revealing differences at cellular levels that prove these patients are not as similar as we were led to believe with traditional imaging technology.




* Retinal dystrophies. In other ocular disease, FAF demonstrates variances that may or may not correlate with color fundus imagery. However, a common theme is that the FAF results are indicative of RPE changes occurring on a molecular level that may be precursors to visible, clinically evident disease progression. For example, the evaluation of a patient with Best's disease demonstrates clinically evident, yellow-orange LF retinal lesions (figures 4a-c). When comparing the appearance of these lesions on color fundus photos to FAF imagery, there is good correlation between the LF deposits (seen on color image) and bright, hyperfluorescent areas (on FAF image). Sub-clinically, however, FAF shows extensive mottling of hypofluorescence in the macula where there is RPE cell death, and additional areas of hyperfluores-cence at the borders of the lesion, suggesting where the disease may progress in this example.



* Glaucoma. FAF patterns in patients with suspected and advanced primary open-angle glaucoma, normal-tension glaucoma, pseudoexfoliative glaucoma and even ocular hypertension have all shown evidence of hyperfluorescence in the parapapillary region of the optic nerve head. In some cases, the amount of hyperfluorescence has been correlated to the severity of the disease, with increasing hyperfluorescence associated with more advanced glaucoma. Histology studies confirm the presence of significant LF accumulation within RPE cells in this region, which may signify degeneration not yet clinically evident.8

* Choroidal lesions. In cases of choroidal lesions, optometrists are trained to look for several factors that help to differentiate nevi from melanomas and determine the likelihood of growth. These include the presence of subretinal fluid, lesion thickness, visual symptoms, proximity to the optic nerve head, and presence of LF on the surface of the choroidal lesion. FAF imagery has been used to assist in the detection of subtle LF, especially in less visible, deep or amelanotic lesions. Serial FAF imagery of the choroidal lesion has also been used to monitor for changes in LF that indicate a possible tumor growth.

The presence of LF may be found on both choroidal nevi and melanomas; however, a nevus more commonly presents with patchy, distinct hyperfluorescent patterns (figures 5a,b), whereas melanomas can present either as patchy or as a diffuse pattern of hyperfluorescence with less distinct borders covering at least 50% of the lesion.

Advanced Posterior Pole Imaging

Autofluorescence technology can be used in conjunction with existing posterior pole imaging techniques to provide a more complete clinical picture. For example, color fundus photography, software-assisted RGB filters, emboss filters and OCT (which are discussed below) all provide valuable information about the overall assessment of each patient case.

For example, RGB filters originate from a raw (untouched by the camera's co-processor and with full pixel resolution) digital fundus-camera color image that is composed of three color channels of varying wavelength: red (25%), green (50%) and blue (25%). When isolated, each filter becomes a further study of ocular structures within a specific layer of the posterior pole.9

By isolating the blue channel (wavelength 490nm to 510nm), the resulting image highlights the superficial nerve fiber layer (NFL) for visualization of dropout, for differentiation of a cotton-wool spot (NFL infarct) from a druse, and to better visualize cup-to-disc ratio.

Isolating the green channel (530nm to 550nm) results in high-contrast imagery that highlights retinal structures including retinal hemorrhages or exudates (differentiated from choroidal drusen). Similar to conventional red-free images, the green layer is most helpful in the assessment of vascular disease such as diabetic retinopathy, or artery and vein occlusions.

Finally, when isolating the red channel (wavelength 590nm to 610nm), the result highlights the choroidal layer and allows for choroidal vasculature, RPE and drusen evaluation. The red layer is helpful when studying AMD or when differentiating a flat nevus limited to the choroid from a thick, growing melanoma that has invaded into the retina.

Additionally, an emboss filter is used to address the lack of depth perception when evaluating single-image, digital retinal photos. While stereo imaging can provide some aspect of depth perception, it involves a learned technique and serial imaging. This is one reason why imagery does not substitute for dilated fundus evaluation in which optometrists can view the fundus in stereo, determine ocular health and document pertinent findings. Rather, imagery enhances the information optometrists gather and allows for concise documentation. It is possible to create an embossed, topographical image from a single-color fundus photo or a single RGB channel image because color images are typically represented by a depth range from eight to 24 bits (figures 6a-c).10 Emboss images create a pseudo-stereo effect with three-dimensional-like representation of elevations or depressions within the patient's posterior pole. The greater the bit depth (z axis) of an image, the higher its resolution and the more information the image yields.

All of the above techniques can be incorporated with image registration patterns and fade-in, fade-out technology, which allows serial images to track any progressive changes over time, and which can be viewed simultaneously over each other to correlate color images, RGB filtered layers, and topographical alterations—all from a single image. By incorporating autofluorescence, one can study underlying posterior pole changes simultaneously—in essence "peeling away" the different layers to investigate what findings lie in which layer and what might possibly become future pathology.

The final piece to advanced posterior pole technology would be the incorporation of high-resolution spectral domain OCT. This noninvasive tool can complete up to 70,000 A-scans per second to create detailed, cross-sectional imagery of the retina, or even anterior segment structures. The OCT has essentially provided the practicing clinician with a microscope capable of high-resolution, histologic views of ocular anatomy in vivo during the course of routine clinical examinations.

One of our main goals as primary eye care physicians is to detect, at the earliest possible point, any degenerative changes that could lead to ocular dysfunction. In an effort to step outside of traditional practice, much of our energy should be directed toward not only the accurate identification, documentation and management of eye disease--which is greatly enhanced with advanced posterior pole technology--but also toward preventative eye care. It is important to educate patients on healthy lifestyle practices and proper nutrition for the best support of ocular health. The use of FAF technology can play an important role in demonstrating pending degenerative changes to the fundus for our patients.

Evidence of changes at the level of the RPE in LF distribution that is demonstrated with FAF imagery has been correlated with early pathology that may not yet be clinically visible. Evidence continues to support the concept that excessive accumulation of LF in the RPE can lead to cellular destruction, retinal aging and visual degeneration. As technologies refine their presence in our practice, we are able to better understand pathology, detect possible at-risk patients earlier, and direct counseling and preventative health care even sooner. This will undoubtedly improve patient care if earlier steps in disease intervention can save or slow the natural progression of ocular disease as we see it today.

Dr. Shahid is clinical assistant professor at the University of Iowa's Carver College of Medicine, Department of Ophthalmology and Visual Science, Iowa City, Iowa.


  1. Terman A, Brunk UT. Lipofuscin. Int J Biochem Cell Biol. 2004 Aug;36(8):1400-4.
  2. Seehafer SS, Pearce DA. You say lipofuscin, we say ceroid: defining autofluorescent storage material. Neurobiol Aging. 2006 Apr;27(4):576-88. Epub 2006 Feb 7.
  3. Jung T, Höhn A, Grune T. Lipofuscin: detection and quantification by microscopic techniques. Methods Mol Biol. 2010;594:173-93.
  4. Yanuzzi LA, Ober MD, Slakter JS, et al. Ophthalmic fundus imaging: today and beyond. Perspective. Am J Ophthalmol. 2004 Mar;137(3):511-24.
  5. Schmitz-Valckenberg S, Fleckenstein M, Scholl HPN, Holz FG. Fundus autofluorescence and progression of age-related macular degeneration. Surv Ophthalmol. 2009 Jan-Feb;54(1):96-117.
  6. Janik-Papis K, Ulinska M, Krzyzanowska A, et al. Role of oxidative mechanisms in the pathogenesis of age-related macular degeneration. Klin Oczna. 2009;111(4-6):168-73.
  7. Holz FG, Bindewald-Wittich A, Fleckenstein M, et al; FAM-Study Group. Progression of geographic atrophy and impact of fundus autofluorescence patterns in age-related macular degeneration. Am J Ophthalmol. 2007 Mar;143(3):463-72.
  8. Laemmer R, Horn FK, Viestenz A, et al. Measurement of auto-fluorescence in the parapapillary atrophic zone in patients with ocular hypertension. Graefes Arch Clin Exp Ophthalmol. 2007 Jan;245(1):51-8.
  9. Szirth B, Khouri A, Bhagat N, Shahid K. New concepts in screening for vision threatening disease (1578/B43). Poster presented at Association for Research in Vision and Ophthalmology meeting, May 7, 2007; Ft. Lauderdale, Fla.
  10. Zarbin MA, Szirth BC. Current treatment of age-related macular degeneration. Optom Vis Sci. 2007 Jul;84(7):559-72.