Differentiable Projection from Optical Coherence Tomography B-Scan without Retinal Layer Segmentation Supervision

TL;DR

This paper introduces a differentiable, end-to-end framework for generating projection maps from OCT B-scans without explicit retinal layer segmentation, leveraging implicit coordinate prediction and pixel interpolation.

Contribution

It proposes a novel Differentiable Projection Module that enables direct training on projection maps, improving quality and offering a new neural representation for geometric deep learning.

Findings

Outperforms baseline models in projection map quality

Eliminates need for explicit retinal layer segmentation

Provides a new neural representation for geometric areas

Abstract

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.