Problem
Research questions and friction points this paper is trying to address.
OCT-to-OCTA synthesis
clinical utility
reconstruction-utility gap
downstream-task fidelity
diabetic retinopathy
In Defense of OCTA: The Reconstruction-Utility Gap in OCT-to-OCTA Synthesis
🤖 AI Summary
This study addresses the discrepancy between high reconstruction metrics and insufficient clinical utility in OCT-to-OCTA synthesis by proposing downstream task performance as a novel evaluation criterion. Through frozen segmenter probing, paired Wilcoxon tests, and matched blur control experiments, we demonstrate that high fidelity does not equate to clinical usability. Results indicate that synthetic images exhibit significantly reduced Dice coefficients for capillaries and fail to reproduce neovascularization. This work establishes and validates downstream task fidelity as a critical evaluation paradigm, revealing the gap between traditional metrics and clinical requirements. Consequently, it provides a more reliable assessment framework for medical image synthesis, emphasizing that clinical efficacy must supersede conventional reconstruction scores in evaluating synthetic data quality.
📝 Abstract
Optical coherence tomography angiography (OCTA) images retinal blood flow, giving capillary-perfusion and foveal-avascular-zone biomarkers that grade diabetic-retinopathy ischemia. Because OCTA hardware is less common than structural OCT, recent work synthesizes it from OCT, reporting strong reconstruction (3D PSNR > 31 dB, SSIM > 0.9). We ask not whether the synthetic image looks similar, but whether it supports the measurements OCTA is acquired for. A frozen real-OCTA segmenter, applied as a probe to two synthesizers (XOCT, TransPro), shows downstream Dice falling with structural fineness: large vessels survive (0.862 -> 0.831) while the fine capillary network collapses (0.798 -> 0.635, five times the large-vessel loss; paired Wilcoxon p < 1e-3), TransPro worse throughout. A matched-blur control shows this detail is fabricated, not blurred. Retrained on a private Spectralis dataset, neither synthesizer reproduces the neovascular lesion (qualitative, n=3). Reconstruction fidelity is not clinical utility; we establish downstream-task fidelity as the evaluation OCT-to-OCTA synthesis needs.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Reconstruction-Utility Gap
Downstream-task Fidelity
OCT-to-OCTA Synthesis
Clinical Utility Evaluation
Capillary Network Collapse
🔎 Similar Papers
No similar papers found.
