Do We Need Fluorescein Angiography? Noninvasive Imaging of the Eye Fundus

Abstract

Fluorescein angiography has contributed much to our present knowledge but it is an invasive method which may be associated with serious complications. It can be replaced today in most diagnostic situations such as vein occlusions, diabetic retinopathy and age-related macular degeneration by the combination of two noninvasive diagnostic procedures: fundus digital photography with computer-aided detection systems and high definition optical coherence tomography.

Copyright © 2012 S. Karger AG, Basel

Fluorescein angiography documents if there is fluorescein leakage which in turn indicates disruption of the blood-retinal barrier (BRB). Clinical use of fluorescein angiography has contributed significantly to the present understanding of retinal disease, particularly to determine alterations of the inner BRB at the retinal vascular level or of the outer BRB, at the retinal pigment epithelium.

The major problem of its use is the need to inject sodium fluorescein, that is used as a tracer, which is a small molecule that diffuses freely through the choriocapillaris and Bruch’s membrane but does not diffuse through the tight junctions of the retinal endothelial cells and the retinal

pigment epithelium. Fluorescein angiography also contributes by identifying well capillary and vessel closure (fig. 1).

Intravenous injection of sodium fluorescein is generally safe and easy to perform. However severe anaphylactic reactions may occur [1] and attention should be given to aging patients particularly if they have a history of cardiac or cerebrovascular ischemic disease.

Between 1986 and 2008 there were 28 reports of death associated with fluorescein angiography in the safety database. Thirteen were attributed to anaphylactic shock but other associated causes were registered for the remaining. It remains a concern particularly nowadays when the patients are followed at older ages and with a variety of comorbidities.

Noninvasive Imaging of the Eye Fundus

The association of digital fundus imaging enhanced by computer-aided detection systems and analysis with optical coherence tomography is expected to offer all the necessary information in the near future. Both methods are noninvasive, complementary and establish the bridge between

simple clinical imaging of the fundus and highdefinition structural and functional information on the choroid and retina.

The combination of two noninvasive methods of fundus imaging, digital fundus photography and optical coherence tomography, has already demonstrated its value by allowing the identification of different diabetic retinopathy phenotypes of progression [2].

Fundus viewing can be documented by fundus photography and is a fundamental source of information regarding the diagnosis and management of retinal diseases and as an observation window to the body and, particularly, brain circulation. It is a simple examination, noninvasive and well tolerated by the patient.

Improvements in digital technology offer unique opportunities to the development of software tools that facilitate data gathering from digital fundus images and their quantification. Computer-aided detection systems algorithms can already detect a variety of retinal lesions using digital retinal images.

Other algorithms are being developed to quantify and measure retinal lesions. Furthermore, algorithms have been developed to allow comparisons, over a sequence of visits, of the changes occurring in the retina, thus evaluating better and more reliably the progression of retinal disease.

Activity and Progression of Retinal Diseases

Color fundus imaging is the most frequent used imaging modality because of being non-invasive, well accepted by patients and, above all, because it allows recording the visible state of the retina at a particular instant in time, both to document in a permanent fashion and to allow for a deeper and extended analysis. To evaluate progression over time, a basic and widely used technique is to evaluate the temporal sequence to acknowledge the visible changes in the retina [3].

Color fundus images undergo a pre-processing stage to normalize for the acquisition conditions while retaining as much as possible information due to any retinal condition.

All images here considered are of the RGB type (red-green-blue channels) either originally digitally taken or digitized from slide transparency films.

To capture changes in the eye fundus between any two images, RGB images are converted to grayscale in a way to preserve all necessary details.

While the green channel has been used traditionally when analyzing fundus images, because of being the one presenting the best contrast among the RGB channels, it is crucial to capture all available information by principle component analysis. By using this method, a gray-scale image is computed encoding information from the three color channels of the RGB image, being the one presenting the maximum contrast.

A mandatory step for automatically comparing imagesistheirco-registration,oneimageagainstthe other. This is to say that one of the images needs to be projected to the image space of the other, which acts as a reference image. Following this procedure, both share a common reference, being possible to establish a direct pixel-to-pixel correspondence.

In order to achieve the required image coregistration, it is necessary to identify eye fundus natural landmarks, intrinsic fiducial marks, and compute the transformation matrix that, applied to one image, will project it to the image space of the reference image.

Two major steps are incorporated in the above concept. One relates to the identification and classification of the fiducial markers, while the other relates to linking similar fiducial markers between any two images to co-register.

A natural source for fiducial markers is the retinal vascular network, an imprint for each human eye. Vessels characteristics, bifurcations and crossovers, allow establishing possible links between any two images from the same eye. After having found the true links for several fiducial markers, one can compute the respective transformation matrix.

After the process of image co-registration and having both gray-scale images sharing a common referential, a pixel-bypixel difference (subtraction) can be computed between these gray-scales images, which represent the visible state of a human retina for two points in time.

This difference produces an image difference summarizing the changes that occurred within the time interval between the two images (fig. 2). It is now possible to identify activity of retinal disease using this methodology.

Optical Coherence Tomography

Optical coherence tomography (OCT) is proving to be an accurate tool for early diagnosis, analysis and monitoring of retinopathy with

high repeatability and resolution. OCT allows not only the qualitative diagnosis of macular edema, but also the quantitative assessment of edema (fig. 3). It identifies the different retinal layers and high-definition OCT is able to identify ganglion cell loss and ischemia, as well as photoreceptors density predicting potential visual recovery.

In the case of macular edema, OCT demonstrates increased retinal thickness with areas of low intraretinal reflectivity [4]. Hard exudates are detected as spots of high reflectivity with low reflective areas behind them.

OCT is particularly useful to detect features such as serous retinal detachment. It appears as a shallow elevation of the retina, with an optically clear space between the retina and the retinal pigment epithelium, and distinct outer border of the detached retina.

OCT is also particularly relevant to analyze the vitreomacular relationship.

Finally one major advantage of OCT is that it allows measurement of retinal thickness from the tomograms by means of computer imageprocessing techniques. OCT allows retinal thickness to be calculated as the distance between the anterior and posterior highly reflective boundaries of the retina, which are located by a thresholding algorithm (fig. 4).

The good reproducibility of retinal thickness measurement with OCT allows its use for longitudinal objective monitoring of macular edema, and for the assessment of treatment efficacy.

Treatment Decisions

Present-day treatment decision on retinal diseases, including central vein occlusions, branch vein occlusion, diabetic retinopathy and age-related macular degeneration can be made only based on fundus digital imaging and OCT.

The hemorrhages, venous abnormalities and retinal edema that characterize the evolution of

Fig. 2. Co-registration of different fundus images of the same eye showing differences in hard exudates between different examinations. A different color is given for each comparison between successive images.

Fig. 3. Identification of areas of increased retinal thickness (edema) in the macula by OCT.

vein occlusions, both central and branch occlusions, can be followed by fundus digital imaging and OCT. Only rarely is fluorescein angiography needed to document ischemia.

The same occurs with diabetic retinopathy where the decisions to treat macular edema

or proliferative diabetic retinopathy are generally addressed without the need for fluorescein angiography.

Finally, AMD diagnosis and treatment by intravitreal injections needs essentially OCT and fundus digital imaging.

References

1 Yannuzzi LA, Rohrer KJ, Tinder LJ, et al: Fluorescein angiography complications survey. Ophthalmology 1986;93:

611–617.

2Ferreira J, Bernardes R, Baptista P, Cunha-Vaz J: Earmarking retinal changes in a sequence of digital color fundus photographs. IFMBE Proc 2005; 11:1727–1983.

3 Nunes S, Pires I, Rosa A, Duarte L, Bernardes R, Cunha-Vaz J: Microaneurysm turnover is a biomarker for diabetic retinopathy progression to clinically significant macular edema: findings for type 2 diabetics with nonproliferative retinopathy. Ophthalmologica 2009;223: 292–297.

4 Cunha-Vaz J, Coscas G: Diagnosis in macular edema. Steroids and management of macular edema. Ophthalmologica 2010;224(suppl 1):2–7.