SAU-Net: Medical Image Segmentation Method Based on U-Net and Self-Attention

被引：0
|
作者：
Zhang S.-J. [1 ]
Peng Z. [1 ]
Li H. [1 ]
机构：
[1] School of Information Science and Technology, Qingdao University of Science and Technology, Shandong, Qingdao
来源：
Tien Tzu Hsueh Pao/Acta Electronica Sinica | 2022年 / 50卷 / 10期
关键词：
decomposition convolution; deep learning; medical image segmentation; self-attention; U-Net;
D O I：
10.12263/DZXB.20200984
中图分类号：
学科分类号：
摘要：
Biomedical image segmentation based on deep learning can better help doctors make an accurate diagno⁃ sis due to its enhanced accuracy. At present, the U-Net-based mainstream segmentation model extracts local features through multi-layer convolutions, which lacks global information and leads to over-localized results with errors. This paper improves the U-Net model through the self-attention mechanism and decomposition convolution and proposes a new deep segmentation network called SAU-Net. The model uses the self-attention module to increase global information, and chang⁃ es the cascade structure in the original U-Net to pixel-by-pixel addition in order to reduce the dimension and cut down the calculation cost. A fast and concise decomposition convolution method is proposed which integrates the traditional convolu⁃ tion into a two-way one-dimensional convolution, and the residual connection is added to enhance the context information. The experimental results conducted on the two brain tumor datasets of BRATS and Kaggle show that SAU-Net has better performance in terms of parameters and the Dice coefficients. © 2022 Chinese Institute of Electronics. All rights reserved.
引用
收藏
页码：2433 / 2442
页数：9
相关论文
共 26 条
  • [1] ALTAF F, ISLAM S M S, AKHTAR N, Et al., Going deep in medical image analysis: Concepts, methods, challenges, and future directions, IEEE Access, 7, pp. 99540-99572, (2019)
  • [2] RONNEBERGER O, FISCHER P, BROX T., U-Net: Con⁃ volutional Networks for Biomedical Image Segmentation, Lecture Notes in Computer Science, pp. 234-241, (2015)
  • [3] ZHOU Z W, SIDDIQUEE M M R, TAJBAKHSH N, Et al., UNet++: A nested U-net architecture for medical image segmentation, Deep Learning in Medical Image Analy⁃ sis and Multimodal Learning for Clinical Decision Sup⁃ port, 11045, pp. 3-11, (2018)
  • [4] KAMNITSAS K, LEDIG C, NEWCOMBE V F J, Et al., Ef⁃ ficient multi-scale 3D CNN with fully connected CRF for accurate brain lesion segmentation, Medical Image Anal⁃ ysis, 36, pp. 61-78, (2017)
  • [5] GHAFOORIAN M, KARSSEMEIJER N, HESKES T, Et al., Non-uniform patch sampling with deep convolutional neural networks for white matter hyperintensity segmenta⁃ tion, 2016 IEEE 13th International Symposium on Bio⁃ medical Imaging, pp. 1414-1417, (2016)
  • [6] HU J, SHEN L, ALBANIE S, Et al., Squeeze-and-excita⁃ tion networks, IEEE Transactions on Pattern Analysis and Machine Intelligence, 42, 8, pp. 2011-2023, (2020)
  • [7] XIE Y P, ZHANG Z Z, SAPKOTA M, Et al., Spatial clock⁃ work recurrent neural network for muscle perimysium seg⁃ mentation, International Conference on Medical Image Computing and Computer-Assisted Intervention, pp. 185-193, (2016)
  • [8] CHEN J X, YANG L, ZHANG Y Z, Et al., Combining fully convolutional and recurrent neural networks for 3D bio⁃ medical image segmentation, NIPS'16: Proceedings of the 30th International Conference on Neural Information Processing Systems, pp. 3044-3052, (2016)
  • [9] TSENG K L, LIN Y L, HSU W, Et al., Joint sequence learn⁃ ing and cross-modality convolution for 3D biomedical seg⁃ mentation, 2017 IEEE Conference on Computer Vision and Pattern Recognition, pp. 3739-3746, (2017)
  • [10] VASWANI A, SHAZEER N, PARMAR N, Et al., Atten⁃ tion is all You need, NIPS'17: Proceedings of the 31st International Conference on Neural Information Process⁃ ing Systems, pp. 6000-6010, (2017)
123