Differentiable Projection from Optical Coherence Tomography B-Scan without Retinal Layer Segmentation Supervision
This work addresses a bottleneck in retinal disease diagnosis by enabling projection map prediction without segmentation supervision, though it is incremental as it builds on existing deep learning methods.
The study tackled the problem of generating projection maps from OCT B-scans without needing supervised retinal layer segmentation, by introducing a differentiable projection module that implicitly represents layers and achieves high-quality results, significantly outperforming baselines.
Projection map (PM) from optical coherence tomography (OCT) B-scan is an important tool to diagnose retinal diseases, which typically requires retinal layer segmentation. In this study, we present a novel end-to-end framework to predict PMs from B-scans. Instead of segmenting retinal layers explicitly, we represent them implicitly as predicted coordinates. By pixel interpolation on uniformly sampled coordinates between retinal layers, the corresponding PMs could be easily obtained with pooling. Notably, all the operators are differentiable; therefore, this Differentiable Projection Module (DPM) enables end-to-end training with the ground truth of PMs rather than retinal layer segmentation. Our framework produces high-quality PMs, significantly outperforming baselines, including a vanilla CNN without DPM and an optimization-based DPM without a deep prior. Furthermore, the proposed DPM, as a novel neural representation of areas/volumes between curves/surfaces, could be of independent interest for geometric deep learning.