FFU-Net: Feature Fusion U-Net for Lesion Segmentation of Diabetic Retinopathy

Abstract

Diabetic retinopathy is one of the main causes of blindness in human eyes, and lesion segmentation is an important basic work for the diagnosis of diabetic retinopathy. Due to the small lesion areas scattered in fundus images, it is laborious to segment the lesion of diabetic retinopathy effectively with the existing U-Net model. In this paper, we proposed a new lesion segmentation model named FFU-Net (Feature Fusion U-Net) that enhances U-Net from the following points. Firstly, the pooling layer in the network is replaced with a convolutional layer to reduce spatial loss of the fundus image. Then, we integrate multiscale feature fusion (MSFF) block into the encoders which helps the network to learn multiscale features efficiently and enrich the information carried with skip connection and lower-resolution decoder by fusing contextual channel attention (CCA) models. Finally, in order to solve the problems of data imbalance and misclassification, we present a Balanced Focal Loss function. In the experiments on benchmark dataset IDRID, we make an ablation study to verify the effectiveness of each component and compare FFU-Net against several state-of-the-art models. In comparison with baseline U-Net, FFU-Net improves the segmentation performance by 11.97%, 10.68%, and 5.79% on metrics SEN, IOU, and DICE, respectively. The quantitative and qualitative results demonstrate the superiority of our FFU-Net in the task of lesion segmentation of diabetic retinopathy.

Automated Detection and Diagnosis of Diabetic Retinopathy: A Comprehensive Survey

“…These lesions have a different detection difficulty which directly affects the performance of the proposed pipeline. Among these lesions, the annotation of MA is more challenging [ 28 , 167 ]. Since this lesion is difficult to detect and is the main sign of DR in early stages, some studies focused on the pixel-wise segmentation of this lesion with DCNNs and achieved high enough scores [ 166 ].…”

CARNet: Cascade attentive RefineNet for multi-lesion segmentation of diabetic retinopathy images

“…In recent years, deep learning algorithms [12][13][14][15][16][17][26][27][28][29] have shown outstanding performance and outperformed traditional methods in lesion segmentation. The existing methods can be classified into two types: encoder-decoder and nonencoder-decoder structure.…”

“…However, Currently, two mainstream ideas exist for solving the task of multi-lesion segmentation. On the one hand, works [12][13][14][15]26] resized or cropped the original fundus images into patches and fed them into the U-Net based network for lesion segmentation. However, they only use a single input and the deconvolution operation cannot better retain the contour detail information of the lesion, resulting in the coarse segmentation result.…”

“…Recently, convolutional neural networks (CNNs) have become widespread in many fields of real life [7][8][9][10][11], numerous deep learning-based methods have been presented for lesion segmentation of DR. The existing methods [12][13][14][15][16][17] for lesion segmentation of DR are categorized into encoderdecoder structures and non-encoder-decoder structures. On the one hand, due to the high resolution of fundus images and GPU memory limitations, works [12][13][14][15] first cropped the original image into patches or resized and input them into U-Net and its variants for lesion segmentation.…”

“…The existing methods [12][13][14][15][16][17] for lesion segmentation of DR are categorized into encoderdecoder structures and non-encoder-decoder structures. On the one hand, due to the high resolution of fundus images and GPU memory limitations, works [12][13][14][15] first cropped the original image into patches or resized and input them into U-Net and its variants for lesion segmentation. However, patch images lack global information, and the deconvolution operation cannot preserve the detailed information of small lesions, such as MA, which makes generating accurate predictions difficult.…”