End-to-End Fovea Localisation in Colour Fundus Images With a Hierarchical Deep Regression Network

Abstract

Accurately locating the fovea is a prerequisite for developing computer aided diagnosis (CAD) of retinal diseases. In colour fundus images of the retina, the fovea is a fuzzy region lacking prominent visual features and this makes it difficult to directly locate the fovea. While traditional methods rely on explicitly extracting image features from the surrounding structures such as the optic disc and various vessels to infer the position of the fovea, deep learning based regression technique can implicitly model the relation between the fovea and other nearby anatomical structures to determine the location of the fovea in an end-to-end fashion. Although promising, using deep learning for fovea localisation also has many unsolved challenges. In this paper, we present a new end-to-end fovea localisation method based on a hierarchical coarse-to-fine deep regression neural network. The innovative features of the new method include a multi-scale feature fusion technique and a selfattention technique to exploit location, semantic, and contextual information in an integrated framework, a multi-field-of-view (multi-FOV) feature fusion technique for context-aware feature learning and a Gaussian-shift-cropping method for augmenting effective training data. We present extensive experimental results on two public databases and show that our new method achieved state-of-the-art performances. We also present a comprehensive ablation study and analysis to demonstrate the technical soundness and effectiveness of the overall framework and its various constituent components.

Bilateral-ViT for Robust Fovea Localization

“…From Table II, the Bilateral-ViT achieves state-of-the-art performance for all the evaluation settings. In particular, on the Messidor dataset, at 1/8R, our network reaches the best accuracy of 85.65% with a gain of 1.84% compared to the second-best score (83.81%) [13]. It also reaches an accuracy of 100% at evaluation thresholds of 1/2R, 1R, and 2R; in other words, the localization errors are at most 1/2R (approximately 19 pixels for an input image size of 512 × 512).…”

“…It consists of 400 images annotated with fovea locations, in which 213 images are pathologic myopia, and the remaining 187 images are normal retinas. For the fairness of comparisons, we keep our data split identical to [13].…”

“…Most deep learning-based methods formulate the fovea localization as a regression task [10]- [13]. Some methods utilize retinal structures, such as OD and blood vessels, as constraints for inferring the location of fovea.…”

Robot Three-Finger Grasping Strategy Based on DeeplabV3+

“…Researchers all over the world aim to make robots achieve intelligent, human-like grasp capabilities, which will expand the application field of robots and create huge economic and social value. With the rapid development of deep learning and the improvement of camera sensor hardware, research on object recognition and location based on machine vision has made great progress [1], but there is less research on object grasp point detection, and it is mainly focused on rectangular grasp strategy [2][3][4][5][6]. Ian Lenz et al [3] proposed a two-layer cascaded deep learning network to predict grasping strategy: The first deep network was used to quickly exclude the impossible grasp options; the second filtered the grasp strategy based on the first network and output the optimal value.…”

Multiscale Softmax Cross Entropy for Fovea Localization on Color Fundus Photography

“…An accurate localization of the fovea, an important anatomical landmark in the retina, can be beneficial to the computer aided diagnosis of retinal diseases. Huang et al [1] take advantage of the geometrical relationship between optic disc and fovea to achieve a more accurate localization and Xie et al [2] utilize a three-stage network with coarse-fine fusion. MSE loss is used in both approaches.…”