Study population

All consecutive patients with AMD with GA, participants of a 2-year prospective controlled trial to assess the effects of rheopheresis treatment on patients with non-exudative AMD (local Ethic Committee approval: ref 2006-A00671-50), who presented at the University Eye Clinic of Creteil between January 2013 and May 2013, underwent several different macular imaging modalities on the same day. Informed consent was obtained from all patients in agreement with the Declaration of Helsinki for research involving human subjects.

Criteria for inclusion were age >55 years, presence of GA on fundus biomicroscopy (unifocal or multifocal) secondary to AMD in at least one eye defined as a lesion measuring at least 175 μm in greatest linear diameter within the macula, with visible choroidal vessels, typically associated to AMD-related phenotypic changes such as hyperpigmentations or soft drusen, with no signs of choroidal neovascularisation, intraretinal or subretinal fluid, or haemorrhage.

Exclusion criteria were, significant media opacities, myopia >6 dioptres, evidence of diabetic retinopathy or any other retinal vascular disease and signs of hereditary retinal dystrophy. Eyes with GA extending outside the arcades were excluded. Before evaluation of GA, image quality for each modality was assessed by one of the authors (GQ); when documented as ‘poor’ (low contrast between atrophic and intact retina in each imaging modality), the corresponding study eyes were excluded from the analysis.

Study protocol

All patients underwent MultiColor, CFP, FAF, NIR-FAF and SD-OCT imaging on the same day.

CFP were obtained using a high-resolution digital retinal camera (TRC-50X, Topcon, Tokyo, Japan; Nikon D2H Camera; field of view 50°×50°; image resolution 1576×1224 pixels).

For high-resolution multimodal fundus imaging, we used a combined instrument (Spectralis SD-OCT, Heidelberg Engineering) that allows for simultaneous recording of cSLO and SD-OCT images.12 By suppressing light from planes anterior and posterior to the plane of interest, cSLO image acquisition results in high-contrast fundus images. Using automated eye tracking and image alignment based on cSLO images, the software allows averaging a variable number of single images in real time (Automatic Real Time Module; Heidelberg Engineering). Individual brightness and contrast settings were optimised after image acquisition. Image brightness was improved by stretching the pixel histogram using the entire range of available pixel values (0–255).15 Image contrast was manually adjusted by subjectively maximally enhancing the visualisation of the details.15 A standardised imaging protocol was performed in all patients, which included acquisition of blue FAF (excitation λ=488 nm; emission λ>500 nm; field of view, 30°×30°; image resolution 768×768 pixels), NIR-FAF (excitation λ=787 nm; emission λ=800 nm; field of view, 30°×30°; image resolution, 768×768 pixels) and MultiColor (a recent imaging mode developed for Spectralis SD-OCT). MultiColor images are composed of three simultaneously acquired reflective images. Each of these images is acquired with a distinctive laser wavelength with cSLO: blue reflectance (486 nm), green reflectance (518 nm) and IR reflectance (815 nm). The final MultiColor image is an overlay of these three images. MultiColor images were acquired in real time and the automatic real-time averaging (up to 100 single images) was used to minimise noise. The ‘standard’ presentation mode was used for the MultiColor images. Owing to chromatic aberration, the tree images composing the MultiColor image all have a slightly different focal plane (∼0.5 to 1 dioptres between IR and blue) and, thus, depending on the focus, distinct retinal structures become more or less visible. To limit potential and important source of error, the focal plane was chosen so that the visualisation of the GA details at the posterior pole was subjectively maximally enhanced. Simultaneous SD-OCT scanning, using a second independent pair of scanning mirrors (λ=870 nm; acquisition speed, 40 000 A-scans per seconds; scan depth, 1.8 mm; digital depth resolution, approximately 3.5 μm per pixel; optical depth resolution, 7 μm; lateral optical resolution, 14 μm), was associated with cSLO imaging.

To test the ability of MultiColor to evaluate GA, we investigated the equivalence of MultiColor, CFP, FAF, NIR-FAF and SD-OCT imaging in measuring atrophic areas and detecting foveal sparing.

Two readers (NBM and AG) measured the total GA area and width through the fovea on the digital images using the built-in measurement software programs (IMAGEnet system; TRC-50IX, Topcon, Tokyo, Japan; Spectralis Acquisition and Viewing Modules; V.5.6.1.0, Heidelberg Engineering, Heidelberg, Germany). Each grader independently made two measurements at two different times. When grading one image, each reader was blind to all other images from the same patient. Using the computer mouse, all GA areas and width through the fovea were manually traced in CFP, FAF, NIR-FAF and Multicolor (figure 1). In CFP and MultiColor, GA was recognised as sharply demarcated areas of partial or complete depigmentation of RPE typically associated with improved visualisation of the choroidal vasculature. Areas corresponding to drusen were not included in the computation. In FAF, and NIR-FAF, GA was recognised as sharply demarcated areas of reduced autofluorescence, generally surrounded by a hyperautofluorescent outer border.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Multimodal imaging in the evaluation of geographic atrophy in the right eye of a patient with age-related macular degeneration and a foveal sparing. The geographic atrophy area and width were measured manually on MultiColor (top left panel), fundus autofluorescence (top middle left panel), near-infrared autofluorescence (top middle right panel) and on colour fundus photography (top right panel) imaging. Spectral-domain optical coherence tomography B scan passing through the fovea (bottom panel) showing the abrupt increase of choroidal signal because of the disruption and loss of photoreceptors and retinal pigment epithelium, which allows the measurement of geographic atrophy width.

The built-in measurement software programs (IMAGEnet system; TRC-50IX, Topcon; Spectralis Acquisition and Viewing Modules; V.5.6.1.0, Heidelberg Engineering) were used to calculate the area of each encircled region by automatic conversion of pixels to square millimetre (based on the magnification factors of the corresponding device). Total GA area for each image was calculated by summing all the areas of GA and subtracting demarcated areas of retinal sparing.

The two readers, at two different times, independently measured the width of GA on horizontal SD-OCT scan cutting the fovea, manually using the built-in measurement software (Spectralis Acquisition and Viewing Modules; V.5.6.1.0, Heidelberg Engineering). The GA limits were identified by choroidal signal enhancement due to RPE loss and excluding areas with foveal sparing (figure 1).

Finally, the two readers independently evaluated at two different times the presence of foveal sparing in GA on MultiColor, CFP, FAF, NIR-FAF and SD-OCT images.

For each study, the two readers also reported on an Excel sheet (Microsoft Office Excel 2010, Microsoft Corporation, Redmond, Washington, USA) the difficulties encountered while analysing GA limits and foveal sparing on MultiColor, CFP, FAF and NIR-FAF images, particularly with respect to the presence of artefacts, xanthophyl pigment or low contrast.

Statistical analysis

Statistical analysis was performed with SAS V.9.2 statistical software (SAS Institute, Inc, Cary, North Carolina, USA). Intergrader and intragrader agreements were determined using intraclass correlation (ICC) coefficient. The mean of two measurements was used for each grader. Student t test for paired data was used for means comparisons. To quantify intergrader and intragrader agreement concerning the foveal involvement, a κ coefficient was computed. A significant p value was <0.05.