跳到主要导航 跳到搜索 跳到主要内容

GLANCE: Combating Label Noise Using Global and Local Noise Correction for Multi-label Chest X-Ray Classification

  • Northwestern Polytechnical University Xian

科研成果: 书/报告/会议事项章节 › 会议稿件 › 同行评审

1 引用 (Scopus)

摘要

Chest X-ray imaging is essential for diagnosing thoracic diseases, with multi-label classification playing a critical role in identifying multiple conditions from a single image. Despite deep neural networks significantly advancing this field, noisy labels extracted from clinical reports pose a significant challenge, undermining the performance of deep models. Several research attempts have been made to address this issue but fail to consider the critical inter-class correlations prevalent in chest X-ray diagnostics. To this end, we propose a Global and Local Noise Correction framework. Our framework comprises a classification backbone and two primary components: a global noise correction module and a local noise correction module. The global noise correction module calculates the noise transition matrix based on the label co-occurrence frequencies and uses the estimated noise transition matrix to reduce the impact of the noisy labels. The local noise correction module treats the temporal ensembling of samples’ historical predictions as the instance-specific pseudo labels, which also serve as the supervision. The proposed framework addresses the shortcomings of existing techniques, i.e., the unreliability of noise transition matrices in the presence of class imbalances and zero co-occurrence frequencies. Comprehensive experimental results demonstrate that our framework surpasses competing methods, showcasing its superior ability to combat label noise and improve multi-label chest X-ray classification accuracy.

源语言英语
主期刊名Uncertainty for Safe Utilization of Machine Learning in Medical Imaging - 6th International Workshop, UNSURE 2024, Held in Conjunction with MICCAI 2024, Proceedings
编辑Carole H. Sudre, Raghav Mehta, Chen Qin, Cheng Ouyang, Marianne Rakic, William M. Wells
出版商Springer Science and Business Media Deutschland GmbH
页194-204
页数11
ISBN(印刷版)9783031731570
DOI
出版状态已出版 - 2025
活动6th International Workshop on Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, UNSURE 2024, Held in Conjunction with MICCAI 2024 - Marrakesh, 摩洛哥
期限: 10 10月 2024 → 10 10月 2024

丛书

姓名Lecture Notes in Computer Science
卷15167 LNCS
ISSN(印刷版)0302-9743
ISSN(电子版)1611-3349

会议

会议6th International Workshop on Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, UNSURE 2024, Held in Conjunction with MICCAI 2024
国家/地区摩洛哥
市Marrakesh
时期10/10/24 → 10/10/24

学术指纹

探究 'GLANCE: Combating Label Noise Using Global and Local Noise Correction for Multi-label Chest X-Ray Classification' 的科研主题。它们共同构成独一无二的学术指纹。

引用此