2K-Fold-Net and feature enhanced 4-Fold-Net for medical image segmentation. (July 2022)

• Record Type: Journal Article
• Authors: Zhang, Yunchu; Dong, Jianfei
• Is Part Of: Pattern recognition. Volume 127(2022)
• Publication Date: 2022-07
• Subjects: 2K-Fold-Net -- EF3-Net -- U-Net -- AFE -- Image segmentation; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2022.108625
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: A 2K-Fold-Net is proposed to generalize U-Net and its stacked or cascaded variants, with K sub-U-Nets and hence the name. The influence of the fold-pair number K on its performance of segmenting medical images is studied. A special case with K=2 is realized and empowered with the feature enhancing functionalities of the attention-aware feature enhancement method, and is thus named as "Enhanced-Feature-4-Fold-Net", or EF3-Net in short. The proposed networks are tested by four medical image segmentation datasets, and show superior performance as compared with some other popular variants of U-Net. Abstract: For segmenting medical images, U-Net has become a popular and effective tool. However, it also has some shortcomings in segmenting fuzzy boundaries and eliminating interferences. Improvements of the original U-Net have been proposed by many authors, resulting in many variants such as MultiResUNet, DoubleU-Net and W-Net. Based on the common characteristics of these structures, we propose in this work a generalized structure by multiplying the folds of a fully convolutional network (FCN) for even more times, and thus name it as "2K-Fold-Net". The more folds in this structure provide more freedoms to create cross links between the neighboring folds. The influence of the fold-pair number K on its performance is also studied. The realizations with K up to 6 are compared to three other variants of cascaded U-Nets using the CVC-ClinicDB dataset. Then the special case "4-Fold-Net" is further empowered with the feature enhancing functionalities recently seen in the attention-aware feature enhancement method. This new net is hence named as "Enhanced-Feature-4-Fold-Net", abbreviated as "EF 3 -Net". Finally, 2K-Fold-Net and EF 3 -Net have been compared with U-Net, SegNet, DoubleU-Net, MultiResUNet and its variants using four challenging medical image datasets. The results have demonstrated that the proposed nets outperform the other variants of U-Net, even with slightly lower amount of parameters. The code is available on: https://github.com/raik7/EF3-Net .