Blend U-Net: Redesigning Skip Connections to Obtain Multiscale Features for Lung CT Images Segmentation

Pengfei Leng; Zhifei Xu; Zhaohui Zhu; Zhiggeng Pan

doi:10.2174/0115734056268487231029154123

Blend U-Net: Redesigning Skip Connections to Obtain Multiscale Features for Lung CT Images Segmentation

Curr Med Imaging. 2023 Nov 6. doi: 10.2174/0115734056268487231029154123. Online ahead of print.

Authors

Pengfei Leng^{1

2}, Zhifei Xu^{1

2}, Zhaohui Zhu^{1

2}, Zhiggeng Pan^{3

2}

Affiliations

¹ School of Public Health, Hangzhou Normal University, Hangzhou, China.
² Institute of VR and Intelligent System, Hangzhou Normal University, Hangzhou, China.
³ School of Artifical Intelligence (School of Future Technology), Nanjing University of Information Science and Technology, Nanjing, China.

PMID: 37936446
DOI: 10.2174/0115734056268487231029154123

Abstract

Background: Lung cancer is a pervasive and persistent issue worldwide, with the highest morbidity and mortality among all cancers for many years. In the medical field, computer tomography (CT) images of the lungs are currently recognized as the best way to help doctors detect lung nodules and thus diagnose lung cancer. U-Net is a deep learning network with an encoder-decoder structure, which is extensively employed for medical image segmentation and has derived many improved versions. However, these advancements do not utilize various feature information from all scales, and there is still room for future enhancement.

Methods: In this study, we proposed a new model called Blend U-Net, which incorporates nested structures, redesigned long and short skip connections, and deep supervisions. The nested structures and the long and short skip connections combined characteristic information of different levels from feature maps in all scales, while the deep supervision learning hierarchical representations from all-scale concatenated feature maps. Additionally, we employed a mixed loss function to obtain more accurate results.

Results: We evaluated the performance of the Blend U-Net against other architectures on the publicly available Lung Image Database Consortium and Image Database Resource Initiative (LIDC-IDRI) dataset. Moreover, the accuracy of the segmentation was verified by using the dice coefficient. Blend U-Net with a boost of 0.83 points produced the best outcome in a number of baselines.

Conclusion: Based on the results, our method achieves superior performance in terms of dice coefficient compared to other methods and demonstrates greater proficiency in segmenting lung nodules of varying sizes.

Keywords: Lung cancer; deep supervision; lung nodules; medical image segmentation; nested structures; skip connection.