Problem
Research questions and friction points this paper is trying to address.
Optical Coherence Tomography
Test-Time Adaptation
Domain Gap
Image Denoising
Automated Analysis
Test-Time Adaptation in Optical Coherence Tomography Using Trajectory-Aligned Time-Independent Flow
🤖 AI Summary
This work addresses the degraded performance of automated analysis on optical coherence tomography (OCT) images acquired by low-cost devices, which suffer from poor image quality. To mitigate domain shift-induced pixel distribution mismatches, the authors propose a novel test-time adaptation method that uniquely integrates trajectory alignment with time-independent flow matching. Specifically, test images are histogram-aligned to a synthetic reference trajectory, and the time conditioning of a generative model is removed to produce high-quality surrogate images. This approach effectively enhances image fidelity without requiring labeled target-domain data and achieves state-of-the-art performance in segmenting key biomarkers associated with two stages of age-related macular degeneration (AMD).
📝 Abstract
Optical coherence tomography (OCT) is essential in ophthalmology, but inconsistent image quality especially in low-cost devices hinders automated analysis. To address this, we introduce a flow-matching-based test-time adaptation method that generates high-quality surrogate images from noisy inputs. Typically, domain gaps between test and training data cause pixel distribution mismatches during the denoising process. We overcome this by matching the test image's histogram to synthetic reference trajectories, successfully aligning the input with expected distributions. Additionally, we remove the network's time conditioning to account for slight deviations in real-world noise distributions. Our approach achieves state-of-the-art performance in segmenting critical biomarkers for two stages of Age-related Macular Degeneration (AMD). Code is available: https://github.com/Veit21/tta-flow.
Innovation
Methods, ideas, or system contributions that make the work stand out.
test-time adaptation
flow matching
optical coherence tomography
domain alignment
time-independent denoising
🔎 Similar Papers
