Retinal Vessels Segmentation Based on Dilated Multi-Scale Convolutional Neural Network

TL;DR

This paper introduces D-Net, a dilated multi-scale convolutional neural network designed for retinal vessel segmentation, effectively capturing global context and small vessel details, outperforming existing methods on multiple datasets.

Contribution

The paper proposes a novel D-Net architecture with dilation convolution and multi-scale feature fusion for improved retinal vessel segmentation.

Findings

D-Net outperforms state-of-the-art methods in accuracy, sensitivity, and specificity.

Dilation convolution enlarges receptive fields without losing spatial resolution.

Multi-scale feature fusion enhances detection of vessels of various sizes.

Abstract

Accurate segmentation of retinal vessels is a basic step in Diabetic retinopathy(DR) detection. Most methods based on deep convolutional neural network (DCNN) have small receptive fields, and hence they are unable to capture global context information of larger regions, with difficult to identify lesions. The final segmented retina vessels contain more noise with low classification accuracy. Therefore, in this paper, we propose a DCNN structure named as D-Net. In the proposed D-Net, the dilation convolution is used in the backbone network to obtain a larger receptive field without losing spatial resolution, so as to reduce the loss of feature information and to reduce the difficulty of tiny thin vessels segmentation. The large receptive field can better distinguished between the lesion area and the blood vessel area. In the proposed Multi-Scale Information Fusion module (MSIF), parallel convolution layers with different dilation rates are used, so that the model can obtain more dense feature information and better capture retinal vessel information of different sizes. In the decoding module, the skip layer connection is used to propagate context information to higher resolution layers, so as to prevent low-level information from passing the entire network structure. Finally, our method was verified on DRIVE, STARE and CHASE dataset. The experimental results show that our network structure outperforms some state-of-art method, such as N4-fields, U-Net, and DRIU in terms of accuracy, sensitivity, specificity, and AUCROC. Particularly, D-Net outperforms U-Net by 1.04%, 1.23% and 2.79% in DRIVE, STARE, and CHASE three dataset, respectively.