Deep Retinal Image Segmentation with Regularization Under Geometric Priors

TL;DR

This paper introduces a domain-enriched deep learning approach for retinal vessel segmentation that incorporates geometric priors and regularization, achieving superior accuracy and efficiency on benchmark datasets.

Contribution

It proposes a novel deep network with geometric feature learning and regularization constraints, improving retinal vessel segmentation performance.

Findings

Outperforms state-of-the-art methods on benchmark datasets.

Achieves higher segmentation accuracy with fewer parameters.

Demonstrates robustness across various training scenarios.

Abstract

Vessel segmentation of retinal images is a key diagnostic capability in ophthalmology. This problem faces several challenges including low contrast, variable vessel size and thickness, and presence of interfering pathology such as micro-aneurysms and hemorrhages. Early approaches addressing this problem employed hand-crafted filters to capture vessel structures, accompanied by morphological post-processing. More recently, deep learning techniques have been employed with significantly enhanced segmentation accuracy. We propose a novel domain enriched deep network that consists of two components: 1) a representation network that learns geometric features specific to retinal images, and 2) a custom designed computationally efficient residual task network that utilizes the features obtained from the representation layer to perform pixel-level segmentation. The representation and task networks are {\em jointly learned} for any given training set. To obtain physically meaningful and practically effective representation filters, we propose two new constraints that are inspired by expected prior structure on these filters: 1) orientation constraint that promotes geometric diversity of curvilinear features, and 2) a data adaptive noise regularizer that penalizes false positives. Multi-scale extensions are developed to enable accurate detection of thin vessels. Experiments performed on three challenging benchmark databases under a variety of training scenarios show that the proposed prior guided deep network outperforms state of the art alternatives as measured by common evaluation metrics, while being more economical in network size and inference time.