Can OCT Fluid Shape Reveal the Leak in CSCR? Lincoff-Inspired Rules From MICRoN Report 15

Angiography identifies focal leakage in CSCR but requires intravenous dye and appropriate equipment. OCT is routine and noninvasive. Inspired by Lincoff’s rules for localizing retinal breaks from fluid patterns in rhegmatogenous detachment, the authors asked whether SRF contours in CSCR reveal a “primary pool” around the leakage point, distinct from secondary macular pooling and inferior gravitational spread. The analogy is conceptual: CSCR and rhegmatogenous retinal detachment have different fluid volumes and mechanisms.

Eligible patients were adults with acute CSCR, one active leakage point on fundus fluorescein angiography (FFA), and baseline OCT volume imaging. Eyes with incomplete superior SRF/leak coverage, inadequate images, relevant coexisting retinal disease, or baseline macular neovascularization were excluded. The investigators analyzed 6 × 6-mm OCT volumes (49 horizontal B-scans) and derived outer-retinal en face OCT maps. When inferior SRF extended beyond the OCT scan, infrared reflectance could assist boundary delineation. FFA and en face OCT were manually aligned by retinal vascular landmarks; SRF outlines were manually traced in ImageJ. The study population included 71 men and 35 women; mean age was 45.4 ± 10.8 years. Ink-blot FFA leakage occurred in 96/107 eyes (89.7%) and smokestack leakage in 11/107 (10.3%).

1. Pattern 1 — extramacular primary pocket: 5 eyes (4.7%), a small localized oval/circular or asymmetric SRF collection away from the fovea.
2. Pattern 2 — extension toward the fovea: 3 eyes (2.8%), an asymmetric, often inverted-pear-shaped pocket extending into the foveal region.
3. Pattern 3 — macular pooling: 67 eyes (62.6%), a larger macular collection, either pear-shaped (34 eyes) or oval (33 eyes; 17 symmetric/near-symmetric, 16 asymmetric).
4. Pattern 4 — inferior gravitational extension: 32 eyes (29.9%), a vertically elongated, pear-like, oval, or dumbbell configuration, sometimes extending beyond the inferior macula.

Figure 2 (page 4) provides FFA–en face OCT examples of each category. Figure 3 (page 5) illustrates oval versus pear-shaped variants and selected eyes changing configurations over time. Figure 8 (page 10) depicts a hypothesized progression from a primary pocket to macular pooling and inferior spread; it is not a validated longitudinal staging sequence.

The researchers treated the extrafoveal bulge of SRF as a putative primary pool, with a second macular pool centered near the fovea. A best-fit circle was positioned along the relevant SRF curvature, and its centroid was taken as the predicted leak coordinate relative to the foveal center. For asymmetric pear-shaped collections, the circle follows the bulge farthest from the fovea. For asymmetric oval collections, it follows the outermost curvature. For symmetric/near-symmetric ovals, the circle is placed against the superior SRF margin. For gravitational configurations, the superior SRF contour is examined, selecting the nonmacular superior bulge when distinct. The OCT predictor and FFA ground-truth coordinates were marked by separate masked graders; a preliminary 20-eye interobserver exercise reported ICC above 0.9. Figure 1 (page 3) demonstrates registration, tracing, and circle placement; Figure 9 (page 12) provides the authors’ decision flowchart. This was a manual, morphology-informed method, not an automated algorithm.

• Mean absolute Euclidean localization error: 0.26 mm; median 0.21 mm; root mean square error 0.32 mm.
• Within 0.25 mm: 60/107 eyes (56.1%); within 0.5 mm: 86/107 (80.4%); within 1 mm: 107/107 (100%).
• Coordinate agreement: ICC 0.97 (95% CI 0.95–0.98) on the horizontal axis and 0.96 (95% CI 0.95–0.98) on the vertical axis.
• Same retinal quadrant: 90/107 (84.1%). Same horizontal or vertical hemisphere: 104/107 (97.2%); each individually defined horizontal and vertical hemisphere agreed in 97/107 (90.6%).

Figure 6 (page 8) shows Bland–Altman analyses, with small mean coordinate biases but limits of agreement spanning approximately 0.4–0.5 mm in either direction. Crucially, the rules were developed and assessed on the same 107 eyes; these are apparent/internal performance estimates, not external validation.

Table 1 (page 8) reports mean errors of 0.07 mm for pattern 1 (n=5), 0.18 mm for pattern 2 (n=3), 0.26 mm for pattern 3 (n=67), and 0.29 mm for pattern 4 (n=32). Within pattern 3, mean error was 0.45 mm for symmetric/near-symmetric oval pools, 0.25 mm for asymmetric ovals, and 0.17 mm for pear-shaped collections. Nonoval/asymmetric patterns were more accurate than symmetric/near-symmetric oval patterns (reported P < 0.001). The very small pattern-1 and pattern-2 subgroups cannot support stable pattern-specific precision estimates. The reported nonoval-pool standard deviation of 1.13 in Table 1 appears internally inconsistent with its mean error of 0.18 and RMSE of 0.22; this value should not be used without author clarification.

The authors propose two principal influences: preferential accumulation toward the fovea, attributed to posterior-pole curvature, and inferior extension under gravity once a macular pool reaches a putative capacity. Macular pooling was observed in 95/107 eyes (88.7%). Spatial overlap of pattern-4 fluid masks yielded a median macular-pool area of 8.23 mm² (major/minor axes approximately 3.8/2.7 mm). A receiver operating characteristic analysis discriminating pattern 4 from pattern 3 using SRF area yielded AUC 0.80 (95% CI 0.70–0.90), with a cohort-derived cutoff of 7.14 mm² (84% sensitivity, 64% specificity). A second ROC analysis relating leak–fovea distance to a symmetric oval configuration reported AUC 0.86 (95% CI 0.75–0.94), with a 1.14-mm threshold (94% sensitivity, 70% specificity). Figure 4 (page 7) displays both ROC curves, and Figure 5 (page 7) shows the SRF-overlap heat map. These thresholds are exploratory and sample-derived, not prospectively validated cutoffs; the proposed forces are physiologic hypotheses rather than directly measured causal mechanisms.

Among 73 eyes with longitudinal observations, 48 (65.7%) showed SRF resolution without enlargement, 9 (12.3%) remained stable, and 16 (21.9%) enlarged or changed pattern. Eight of those 16 increased in size alone; eight increased in size and changed configuration. Five transitioned from pattern 3 to pattern 4; one each showed pear-to-oval change, pattern 1 to 2, and pattern 2 to 3. These observations are compatible with selected morphologic transitions but do not establish a universal sequential natural history, especially because many eyes resolve without traversing all four patterns.

En face OCT fluid geometry may provide a probabilistic clue to where an active CSCR leak lies when FFA is unavailable or contraindicated, particularly for a distinctly asymmetric collection. It might also help focus review of angiographic images. However, the study did not prospectively test OCT-guided treatment selection, visual outcomes, safety, or noninferiority to angiography. A mean error of 0.26 mm may be too large for safe focal laser placement near the fovea. The authors suggest that the wider treatment field used in photodynamic therapy (PDT) might tolerate this error, but this is a clinical hypothesis, not evidence that FFA or indocyanine green angiography can be omitted when planning PDT. The proposed rule assumes one active leak; it has not been validated for multifocal, recurrent, or chronic CSCR, or eyes with macular neovascularization.

• Strength: anatomically intuitive hypothesis, carefully defined fluid phenotypes, masked marking of predicted and angiographic leak coordinates, and quantitative spatial error metrics.
• Derivation and validation overlap: performance is likely optimistic because rules were created and tested on the same dataset; no independent temporal, geographic, or external test set.
• Selection and spectrum limitations: only acute, single-leak, adequately imaged cases; results cannot be generalized to multifocal or chronic disease.
• Manual workflow: OCT–FFA registration, segmentation, and best-fit circle placement require operator judgment; preliminary grader agreement does not replace external reproducibility testing.
• Small subgroups: patterns 1 and 2 comprised only five and three eyes.
• Mechanistic uncertainty: centripetal and gravitational vectors were inferred from contours rather than directly measured; the proposed four-step progression remains hypothetical.
• Clinical utility unproven: no trial established treatment-targeting accuracy, reduced angiography use, visual benefit, or avoidance of adverse laser placement.

MICRoN Report 15 demonstrates that SRF shape on en face OCT contains useful spatial information about an angiographically confirmed single CSCR leak. Its apparent 0.26-mm mean localization error is encouraging, especially in asymmetric patterns, but same-cohort rule derivation, manual processing, and restriction to acute single-leak eyes prevent the method from replacing angiographic localization today. Prospective, masked external validation and treatment-specific safety studies are needed before clinical deployment.