Label Decoupling and Reconstruction: A Two-Stage Training Framework for Long-tailed Multi-label Medical Image Recognition

doi:10.1145/3664647.3680606

科研成果详情

题名	Label Decoupling and Reconstruction: A Two-Stage Training Framework for Long-tailed Multi-label Medical Image Recognition
作者	Huang, Jie 1; Chen, Zhao-Min 1; Zhang, Xiaoqin 1; Ge, Yisu 1; Ye, Lusi2; Zhang, Guodao 3; Chen, Huiling 1
会议录名称	MM 2024 - Proceedings of the 32nd ACM International Conference on Multimedia 影响因子和分区
语种	英语
原始文献类型	Conference article (CA)
会议名称	32nd ACM International Conference on Multimedia, MM 2024
会议日期	October 28, 2024 - November 1, 2024
会议地点	Melbourne, VIC, Australia
关键词	Deep learning Diseases Medical image processing Decouplings Gaussian distribution reconstruction Gaussians Label decoupling Long tail Long-tailed distributions Medical image recognition Multi-labels Semantics Information Training framework
摘要	Deep learning has made significant advancements and breakthroughs in medical image recognition. However, the clinical reality is complex and multifaceted, with patients often suffering from multiple intertwined diseases, not all of which are equally common, leading to medical datasets that are frequently characterized by multi-labels and a long-tailed distribution. In this paper, we propose a method involving label decoupling and reconstruction (LDRNet) to address these two specific challenges. The label decoupling utilizes the fusion of semantic information from both categories and images to capture the class-aware features across different labels. This process not only integrates semantic information from labels and images to improve the model's ability to recognize diseases, but also captures comprehensive features across various labels to facilitate a deeper understanding of disease characteristics within the dataset. Following this, our label reconstruction method uses the class-aware features to reconstruct the label distribution. This step generates a diverse array of virtual features for tail categories, promoting unbiased learning for the classifier and significantly enhancing the model's generalization ability and robustness. Extensive experiments conducted on three multi-label long-tailed medical image datasets, including the Axial Spondyloarthritis Dataset, NIH Chest X-ray 14 Dataset, and ODIR-5K Dataset, have demonstrated that our approach achieves state-of-the-art performance, showcasing its effectiveness in handling the complexities associated with multi-label and long-tailed distributions in medical image recognition. © 2024 ACM.
出版者	Association for Computing Machinery, Inc
页码	2861-2869
DOI	10.1145/3664647.3680606
会议录编者/会议主办者	ACM SIGMM
收录类别	EI
发表日期	2024-10-28
EI入藏号	20244817417711
EI主题词	Gaussian distribution
EI分类号	102.1.2 ; 1101.2.1 ; 1106.3.1 ; 1202.1 ; 1202.2
引用统计
文献类型	会议论文
条目标识符	https://kms.wmu.edu.cn/handle/3ETUA0LF/223717
专题	附属第一医院
通讯作者	Chen, Zhao-Min; Zhang, Xiaoqin
作者单位	1.Wenzhou University, Wenzhou, China 2.The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, China 3.Hangzhou Dianzi University, Hangzhou, China
推荐引用方式 GB/T 7714	Huang, Jie,Chen, Zhao-Min,Zhang, Xiaoqin,et al. Label Decoupling and Reconstruction: A Two-Stage Training Framework for Long-tailed Multi-label Medical Image Recognition[C]//ACM SIGMM:Association for Computing Machinery, Inc,2024:2861-2869.