Localizing Anatomical Landmarks in Ocular Images using Zoom-In Attentive Networks

TL;DR

This paper introduces ZIAN, a novel zoom-in attentive network that improves anatomical landmark localization in ocular images by leveraging multi-scale contextual features and an attentive fusion mechanism, outperforming existing methods.

Contribution

The paper proposes a new zoom-in attentive network with multi-scale feature fusion for precise landmark localization in challenging ocular images, addressing the lack of prominent features.

Findings

ZIAN outperforms state-of-the-art localization methods on two ocular image datasets.

The multi-scale, attention-based fusion improves localization accuracy.

The approach effectively handles landmarks with subtle visual features.

Abstract

Localizing anatomical landmarks are important tasks in medical image analysis. However, the landmarks to be localized often lack prominent visual features. Their locations are elusive and easily confused with the background, and thus precise localization highly depends on the context formed by their surrounding areas. In addition, the required precision is usually higher than segmentation and object detection tasks. Therefore, localization has its unique challenges different from segmentation or detection. In this paper, we propose a zoom-in attentive network (ZIAN) for anatomical landmark localization in ocular images. First, a coarse-to-fine, or "zoom-in" strategy is utilized to learn the contextualized features in different scales. Then, an attentive fusion module is adopted to aggregate multi-scale features, which consists of 1) a co-attention network with a multiple regions-of-interest (ROIs) scheme that learns complementary features from the multiple ROIs, 2) an attention-based fusion module which integrates the multi-ROIs features and non-ROI features. We evaluated ZIAN on two open challenge tasks, i.e., the fovea localization in fundus images and scleral spur localization in AS-OCT images. Experiments show that ZIAN achieves promising performances and outperforms state-of-the-art localization methods. The source code and trained models of ZIAN are available at https://github.com/leixiaofeng-astar/OMIA9-ZIAN.