Large Pigment Epithelial Detachments Associated With Type 1 Macular Neovascularization: Do Swept-Source OCT Angiography Devices Detect the Same Thing?

Purpose: Large pigment epithelial detachments (PEDs) compromise automated swept-source optical coherence tomography angiography (SS-OCTA) segmentation, leading to false-negative type 1 macular neovascularization (MNV) in neovascular age-related macular degeneration. We compared the diagnostic sensitivity of two SS-OCTA platforms from different generations to assess whether technological advances offset segmentation limitations. Methods: This was a prospective comparative study of 82 eyes with type 1 MNV and PED height ≥ 250 µm. Each eye underwent dual SS-OCTA imaging with a first-generation 1050-nm (100-kHz) system and a next-generation 1060-nm (400-kHz) system. Two masked readers evaluated automated versus manually adjusted segmentation slabs across PED height categories. Results: On the 1050-nm system, detection improved from 22.0% with automated segmentation to 81.7% after manual adjustment (P < 0.001). The 1060-nm system showed higher automated detection (68.3%), with a modest increase to 75.6% after manual segmentation. Stratified analysis showed superior automated performance of the 1060-nm device in shallow PEDs, whereas intermediate PEDs were most challenging for both systems. In very large PEDs, manual segmentation yielded similar detection rates on both devices. Conclusions: Detection of type 1 MNV beneath large PEDs is significantly device dependent. Although next-generation SS-OCTA systems improve automated detection, manual segmentation remains crucial for optimal sensitivity and accuracy. Device-specific protocols integrating routine manual review are warranted to avoid false-negative findings and treatment delays. Translational Relevance: Manual segmentation remains essential to prevent false-negative detection of type 1 MNV beneath large PEDs, even with next-generation SS-OCTA devices. This has direct implications for diagnostic protocols and treatment decisions in nAMD.