Series Vol. 5 , 25 May 2023
* Author to whom correspondence should be addressed.
Accurate segmentation for Functional Tissue Units (FTUs) is a challenging issue in past decades. In this study, a model using the dataset of tissue section images will be built to evaluate and mark FTUs across five human organs as clearly as possible. We have the Human Protein Atlas (HPA) as training data and the data from Human BioMolecular Atlas Program (HuBMAP) as testing data. To balance accuracy and inference speed, this study applied Unext, an efficient network based on Unet, as the basic model. We also aim to use some tricks to further improve the performance of the model. First, we used several image enhancement methods to diversify the input image. Second, several structures like Feature Pyramid Network (FPN) and the Atrous Spatial Pyramid Pooling are added to improve model performance and convergence speed. As a result, we successfully segment functional tissue units among images of different sizes. Our proposed model scored 0.56 out of 1.00 by the judge of the competition.
machine learning, image segmentation, Unext.
1. Hesamian M et al. 2019 Deep learning techniques for medical image segmentation: achievements and challenges Journal of digital imaging 32(4) 582-596
2. Ronneberger O et al. 2015 October U-net: Convolutional networks for biomedical image segmentation In International Conference on Medical image computing and computer-assisted intervention (pp. 234-241) Springer Cham
3. Amin J et al. 2018 Big data analysis for brain tumor detection: Deep convolutional neural networks. Future Generation Computer Systems, 87, 290-297.
4. Zhang M et al. 2018 RBC semantic segmentation for sickle cell disease based on deformable U-Net. In International Conference on Medical Image Computing and Computer-Assisted Intervention (pp. 695-702) Springer Cham
5. Xu G et al. 2021 Levit-unet: Make faster encoders with transformer for medical image segmentation arXiv preprint arXiv:2107.08623
6. Lin T Y 2017 Feature pyramid networks for object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 2117-2125).
7. Chen L C 2017 Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution and fully connected crfs IEEE transactions on pattern analysis and machine intelligence 40(4) 834-848
8. Song T 2019 U-next: A novel convolution neural network with an aggregation u-net architecture for gallstone segmentation in ct images IEEE Access 7 166823-166832
9. Yun D 2022 ELU-Net: An Efficient and Lightweight U-Net for Medical Image Segmentation IEEE Access vol.10 pp.35932-35941
10. Pontén F et al. 2008 The Human Protein Atlas—a tool for pathology The Journal of Pathology: A Journal of the Pathological Society of Great Britain and Ireland 216(4) 387-393
11. Lin S et al. 2019 The human body at cellular resolution: the NIH Human Biomolecular Atlas Program.
The datasets used and/or analyzed during the current study will be available from the authors upon reasonable request.