Diagnostic Performance of Ring Aperture Retro Mode Imaging for Detecting Pigment Migration in Age-Related Macular Degeneration (AMD)

1. Introduction

Pigment migration in age-related macular degeneration (AMD) is an important clinical finding, reflecting retinal pigment epithelium (RPE) dysfunction and signaling risk of progression toward geographic atrophy (GA) and vision loss [1]. Along with drusen, pigmentary changes are central criteria for upgrading disease classification from early to intermediate AMD (iAMD) [1]. According to the Beckman/AREDS2 consensus classification, iAMD is defined by the presence of large drusen (>125 µm) and/or any AMD-related pigmentary abnormalities in the absence of GA or neovascularization [2].

Pigmentary abnormalities encompass both hyperpigmentation—reflecting RPE migration, hyperreflective foci (HRF), or thickened RPE (tRPE)—and hypopigmentation, which may correspond to early RPE attenuation, nascent GA, or emerging transmission defects on OCT [3]. These changes occur within the tightly coupled RPE–photoreceptor unit. Although RPE dysfunction can compromise photoreceptor survival [4,5], recent OCT-based analyses of AMD progression indicate that photoreceptor alterations with ellipsoid-zone attenuation may also extend beyond or precede detectable RPE loss in early GA, highlighting a bidirectional and stage-dependent relationship between these layers [6].

Multimodal imaging has improved detection of pigment migration. Color fundus images (CFIs) may reveal visible pigment clumping, while structural and en face OCT localize abnormalities as HRF or tRPE [3]. Hyporeflective clumps (HRC) are another lesion detectable with adaptive optics [7,8]. These abnormalities may occur over drusen, at atrophic margins, or on the surface of pigment epithelial detachments (PED). En face OCT can further delineate pigment migration associated with refractile drusen or cuticular drusen [9,10,11].

Fundus autofluorescence (FAF) demonstrates pigment redistribution through characteristic patterns [12], the main signal sources for FAF being RPE lipofuscin and melanolipofuscin [13]. Because the relative contribution of these fluorophores depends on excitation wavelength, conventional visible-light FAF predominantly reflects lipofuscin. Melanolipofuscin, however, exhibits broader emission properties and can generate near-infrared autofluorescence under long-wavelength excitation [14].

Infrared retroillumination (“Retro mode”) was first introduced on the NIDEK F-10 platform and later integrated into the Mirante® multimodal device [15,16]. The system offers three illumination apertures: deviated left (DL), deviated right (DR), and ring aperture (RA). DL and DR configurations create pseudo–three-dimensional shading that accentuates drusen contours, pigment epithelial detachments, and atrophic margins, making them widely used for structural assessment in AMD [17,18,19,20].

In contrast, the ring aperture configuration (RAR) eliminates lateral shadowing and produces a “dark-field” transillumination pattern. Although less visually striking than DL/DR imaging, RAR preferentially highlights small variations in pigment density. When acquisition parameters—particularly flash intensity—are carefully optimized, RAR may reveal subtle pigment redistribution that is difficult to detect on CFIs or FAF, especially near the fovea where macular pigment absorption complicates interpretation.

Despite this potential, the diagnostic value of RAR for detecting pigment migration in AMD has not, to the best of our knowledge, been systematically evaluated. The present study aimed to assess the performance of RAR for detecting pigment migration across different AMD stages and to compare its yield with that of color fundus imaging, FAF, and en face OCT in a real-world, high-volume clinical setting.

2. Materials and Methods

2.1. Study Design

This retrospective observational study was conducted at the Centre de Rétine Médicale, a tertiary referral center for retinal diseases. The study complied with the principles of the Declaration of Helsinki. Written informed consent was obtained from all participants, authorizing the use of their clinical data for research purposes. Ethical approval was obtained from the ethical committee of the French Society of Ophthalmology (SFO).

2.2. Patients

We included consecutive patients examined between May 2023 and February 2025 with a diagnosis of AMD and pigment migration detected on at least one imaging modality by two retina specialists (TD, GL-G). Exclusion criteria were insufficient image quality or the absence of one or more imaging modalities. Based on multimodal imaging characteristics, eyes were categorized as intermediate AMD (iAMD), geographic atrophy (GA), or neovascular AMD (nAMD). Eyes with both GA and nAMD were classified into the nAMD group.

Visual acuity was recorded to provide clinical context but was not used as an outcome measure because pigment migration is not always directly correlated with central visual function, particularly in the presence of GA or media opacity. Refraction was not used analytically because mild ametropia does not affect detection of pigment migration; however, refractive status is reported in the Results section for completeness. The purpose of this study was to evaluate imaging detectability rather than functional impact.

2.3. Imaging Protocol

All imaging data were anonymized before analysis. Examinations were performed after pharmacologic pupil dilation. All images were acquired by a retina specialist (TD), ensuring consistent optimization of acquisition parameters across modalities.

Multimodal imaging included color fundus images (CFIs), fundus autofluorescence (FAF), and Retro mode imaging (DL, DR, and RA) acquired with the Mirante® platform (NIDEK, Japan; 89° field of view), and spectral-domain OCT using the Cirrus HD-OCT 5000 (Carl Zeiss Meditec, USA; axial resolution ~5 µm; acquisition speed 27,000 A-scans/second).

CFIs were obtained with carefully adjusted flash intensity to avoid overexposure, particularly in eyes with GA. Retro mode imaging included DL, DR, and RA scans; however, only RA (RAR) images were analyzed in this study. Flash intensity was progressively reduced to balance contrast and allow visualization of both choroidal vessels and pigment migration. FAF was performed using green excitation (532 nm), minimizing glare and reducing macular pigment absorption.

2.4. En face OCT Acquisition

En face OCT data were derived from 6 × 6 mm scans centered on the fovea. Two Cirrus preset slabs were analyzed:

1 - Retinal IS/OS slab

Used to detect intraretinal hyperreflective foci (HRF), which appear as bright focal dots at the photoreceptor integrity line.

2 - Sub-RPE/choroidal slab

Located beneath the RPE/Bruch’s complex, used to evaluate transmission defects.

• Hypotransmission defects (hypoTD): dark foci caused by signal blockage from thickened RPE (tRPE) or large HRF.
• Hypertransmission defects (hyperTD): bright patches corresponding to RPE attenuation or early atrophy.

For all en face abnormalities, the corresponding B-scans were reviewed to determine whether the lesion represented HRF, tRPE, or RPE loss. This approach follows established en face slab–dependent behavior described in the literature [3,11].

These two slabs were used consistently across all eyes.

2.5. Image Analysis

All images were independently reviewed by two retina specialists (TD and GL-G). Both graders were masked to each other’s findings. After independent assessment, discrepancies were resolved by consensus. For each eye, graders classified AMD stage (iAMD, nAMD, or GA) and scored the presence of pigment migration in each imaging modality as “0” (absent) or “1” (present).

2.5. Statistical Analysis

Intergrader agreement for binary variables (presence of pigment migration on each imaging modality) was assessed using Cohen’s κ coefficient. Agreement for AMD stage classification (iAMD, nAMD, GA) was evaluated using weighted κ. Kappa values were interpreted using Landis and Koch criteria (κ < 0.20: slight; 0.21–0.40: fair; 0.41–0.60: moderate; 0.61–0.80: substantial; >0.80: almost perfect agreement).

To compare diagnostic performance across imaging modalities, en face OCT was used as the reference standard. For CFIs, FAF, and RAR, the following metrics were calculated from 2×2 contingency tables:

• Sensitivity = TP / (TP + FN)
• Specificity = TN / (TN + FP)
• Positive predictive value (PPV) = TP / (TP + FP)
• Negative predictive value (NPV) = TN / (TN + FN), when applicable
• F1 score, defined as the harmonic mean of sensitivity and PPV:F1 = 2 × (Sensitivity × PPV) / (Sensitivity + PPV)

The F1 score was included because it provides a balanced measure of detection performance, particularly when false positives and false negatives have different implications across modalities.

Continuous variables (e.g., visual acuity) are reported as mean ± standard deviation. Categorical variables are presented as counts and percentages. All statistical analyses were performed using Microsoft Excel 365, which is appropriate for agreement coefficients and contingency-based diagnostic performance metrics in retrospective observational studies. A two-sided significance threshold of p < 0.05 was used..

3. Results

A total of 80 eyes from 61 patients were included. The mean age was 79.1 ± 9.8 years, and 39 patients were women. Based on multimodal imaging, 36 eyes were classified as iAMD, 23 as GA, and 21 as nAMD. Eyes with both GA and nAMD were grouped under nAMD. The mean best-corrected visual acuity (BCVA) was 0.59 ± 0.26 decimal units (approximately 20/34 Snellen; mean logMAR 0.28 ± 0.24). Available refraction data indicated mild ametropia, with spherical equivalent values approximately between –3.00D and +2.00D; no high myopia or high hyperopia were present. As expected in a cohort including both iAMD and GA, visual acuity showed variability and was not used as an outcome measure.

In all eyes, en face OCT findings were interpreted by correlating lesions on both slabs with the corresponding B-scans to differentiate HRF from tRPE and to confirm the nature of transmission defects.

Intergrader agreement was high for both AMD stage classification (weighted κ = 0.93) and for the detection of pigment migration across modalities: CFIs (κ = 0.924), FAF (κ = 0.825), en face OCT (κ = 0.902), and RAR (κ = 1.000). RAR demonstrated performance comparable to en face OCT and frequently outperformed CFIs and FAF. Agreement between RAR and en face OCT was high (κ = 0.71), corresponding to a sensitivity of 94.7% and a positive predictive value of 93.4%. Because en face OCT identified pigment migration in most eyes, true negatives were infrequent, resulting in low calculated specificity (true negative in Table 1). In contrast, agreement between RAR and CFIs was moderate (κ = 0.56), with a sensitivity of 91.5% but a lower positive predictive value of 56.6%. These comparisons are summarized in Table 1 and Table 2.

Distinct morphologic patterns of pigment migration were identified on RAR when compared with OCT (Table 3).

On RAR imaging, pigment migration corresponding to HRF on B-scan OCT appeared as thin dark contrast dots or mid-sized dark contrast halos (Figure 1 and Figure 2). Areas of thickened RPE (tRPE) were visualized as broader dark contrast halos on RAR (Figure 3). Refractile drusen, which showed a pyramidal contour with a central hypertransmission core on OCT B-scans, appeared on RAR as ring-shaped dark contrast halos with a bright central core (Figure 4). Cuticular drusen were visible on RAR as small clustered lesions with central dark halos and peripheral light rings (Figure 5).

Compared with CFIs and FAF, RAR frequently provided clearer and more detailed visualization of pigment migration. Pigment clumps tended to appear more extensive and granular, sometimes with a dust-like distribution that was far less apparent on CFIs (Figure 1 and Figure 5). Media opacities commonly reduced the reliability of CFIs and FAF; in contrast, RAR and en face OCT generally preserved diagnostic utility. Most eyes with nAMD had previously received anti-VEGF therapy and showed no active exudation. In two eyes with persistent shallow serous detachments, pigmentary changes were masked on CFIs and FAF but remained visible on RAR and OCT.

In four eyes with GA, subtle pigment migration detected on RAR preceded subsequent foveal involvement by at least six months, later confirmed by en face OCT and FAF (Figure 5). This raises the possibility that RAR may help identify early pigment redistribution with potential prognostic relevance. In one eye with iAMD and cuticular drusen, FAF provided superior visualization of pigment changes compared with RAR, CFIs, or OCT, underscoring the complementary value of multimodal imaging in selected cases.

Top row (September 2022), VA 20/100:

(A) CFI, (B) FAF, (C) RAR, and (D) en face OCT (retinal IS/OS slab) with enlarged corresponding B-scan show early subfoveal pigment migration, including a refractile drusen.In panel (C), white arrowheads indicate thin pigment migration that was detectable on RAR in 2022. In addition to soft drusen, cuticular drusen at the temporal posterior pole appear on RAR as small clusters with central dark halos and peripheral light rings (blue arrowheads in C).The white dotted square in (C) marks the region shown in the en face OCT slab and B-scan in (D).

Bottom row (September 2024), VA 20/250:

(E–H) demonstrate progression to central geographic atrophy. The pigment migration previously highlighted by white arrowheads in panel (C) preceded subsequent foveal atrophy. In addition to soft drusen, cuticular drusen at the temporal posterior pole appear on RAR as small clusters with central dark halos and peripheral light rings (blue arrowheads in C). The white dotted square in (G) again indicates the area displayed in the en face OCT slab and B-scan in (H).

Abbreviations: CFI, color fundus image; FAF, fundus autofluorescence; GA, geographic atrophy; OCT, optical coherence tomography; RAR, ring aperture Retro mode; VA, visual acuity.

4. Discussion

In this retrospective study, RAR imaging proved to be a valuable modality for detecting pigment migration across different stages of AMD. Its sensitivity was comparable to en face OCT, the current reference method, and greater than that of CFIs or FAF, particularly for early pigmentary changes. The ability to achieve rapid acquisition and straightforward interpretation supports its integration into multimodal imaging protocols, especially in high-volume clinical settings.

Pigmentary changes are recognized biomarkers of AMD progression, reflecting RPE dysfunction and an increased risk of conversion to GA or neovascular disease [1,4,5]. These lesions are often subtle and difficult to detect on CFIs or FAF, particularly in the fovea where macular pigment can obscure abnormalities [23]. Although green autofluorescence can reduce the influence of macular pigment absorption [13,24], our findings indicate that Retro mode is more sensitive for detecting pigment migration near the fovea. One likely explanation is that FAF signals in GA arise from multiple mechanisms beyond lipofuscin and melanolipofuscin accumulation [13,24], whereas Retro mode relies primarily on pigment density, with infrared illumination bypassing macular pigment interference. As a result, RAR consistently revealed pigment migration that was underestimated or missed by these modalities. Figure 1 and Figure 5 illustrate how RAR highlights thin or granular pigment patterns not visible on CFIs, and Table 1 and Table 2 demonstrate the superior diagnostic performance of RAR relative to CFIs and FAF. This advantage was particularly evident in eyes with media opacities, where image quality was degraded for other modalities.

Morphologic patterns identified with RAR provide additional insights into the underlying pathophysiology. HRF, well-established OCT biomarkers of progression [25,26,27], appeared on RAR as thin dark contrast dots or mid-sized dark contrast halos (Figure 1 and Figure 2). Thickened RPE, associated with GA progression and basal laminar deposits [3,28,29,30], appeared as broader dark contrast halos (Figure 3). Refractile drusen, typically pyramidal on OCT with central hypertransmission [10,11], were seen on RAR as ring-shaped dark contrast halos with a bright central core (Figure 4). Cuticular drusen displayed small clustered lesions with central dark halos and peripheral light rings (Figure 5), consistent with prior multimodal imaging reports [9,11]. These findings, summarized in Table 3, align with histopathologic and imaging studies highlighting the complexity of RPE–photoreceptor interactions [4,10,31,32].

RAR also provided clinically meaningful information during follow-up. In four eyes with GA, pigment migration detected on RAR preceded foveal involvement by at least six months, later confirmed by OCT and FAF (Figure 5). These findings raise the possibility that RAR may help identify early pigment redistribution with potential prognostic relevance for central atrophy progression. Similarly, in two cases of nAMD with persistent shallow subretinal detachment, pigment migration was obscured on CFIs and FAF but remained visible on both RAR and OCT. Chronic subretinal fluid in nAMD can produce hyperautofluorescent deposits that mask pigment redistribution on FAF [33], whereas RAR retained diagnostic value in this setting.

Cuticular drusen exhibited heterogeneous features across modalities. FAF most often displayed a ring-like pattern, whereas RAR and OCT revealed distinct ring configurations that may correspond to different stages of pigment dispersion or drusen remodeling. These observations underscore the complementary value of multimodal imaging and suggest that RAR may reveal aspects of lesion morphology not fully captured by other techniques.

This study has limitations, including the modest sample size and retrospective design. Although graders were not masked to clinical diagnosis, image interpretation was performed independently and masked to the other grader’s results, with high inter-grader agreement supporting reliability. RAR also requires stable fixation and is susceptible to motion artifacts. Furthermore, although we identified distinct morphologic patterns, the study was not designed to validate whether RAR can reliably differentiate HRF, tRPE, and other lesion subtypes. While RAR provides rapid and intuitive visualization of pigment migration, en face OCT interpretation requires the systematic evaluation of multiple B-scans and slabs, increasing reading complexity. This difference may partly explain the higher ease of use reported with RAR, although OCT remains the reference standard. Despite these limitations, the consistency of findings across modalities supports the clinical utility of RAR as part of a streamlined multimodal approach for AMD.

In conclusion, RAR imaging offers a rapid, practical, and sensitive method for detecting pigment migration in AMD. By complementing OCT and other established modalities, it enhances early recognition of pigmentary changes, supports prognostic assessment, and may provide added value in routine follow-up

5. Conclusions

Ring aperture Retro mode imaging provides sensitivity comparable to optical coherence tomography and superior detection compared with fundus photography and autofluorescence, offering a practical adjunct for early monitoring of pigment migration in age-related macular degeneration.