Scholarlink

Accurate lung cancer diagnosis is crucial for effective treatment and improved outcomes. This study introduces TS-CATMA (Time Series Classification with Adversarial Training and Multi-scale Attention), a novel method designed for lung cancer detection using electronic nose data. TS-CATMA leverages a multi-scale attention mechanism and adversarial training to extract discriminative, domain-invariant features from raw time series data. Evaluated on a lung cancer electronic nose dataset, TS-CATMA achieved a detection accuracy of 90.59% with rapid training (6.15 s) and testing (39.57 ms) times, indicating its potential for early diagnosis. The source code is available at https://github.com/CQU-3DTEAM/TS-CATMA .

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

TS-CATMA: A Lung Cancer Electronic Nose Data Classification Method Based on Adversarial Training and Multi-scale Attention

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Yuze Chen,
  • Lin Yi,
  • Shidan Wang,
  • Fengchun Tian,
  • Ran Liu

摘要

Accurate lung cancer diagnosis is crucial for effective treatment and improved outcomes. This study introduces TS-CATMA (Time Series Classification with Adversarial Training and Multi-scale Attention), a novel method designed for lung cancer detection using electronic nose data. TS-CATMA leverages a multi-scale attention mechanism and adversarial training to extract discriminative, domain-invariant features from raw time series data. Evaluated on a lung cancer electronic nose dataset, TS-CATMA achieved a detection accuracy of 90.59% with rapid training (6.15 s) and testing (39.57 ms) times, indicating its potential for early diagnosis. The source code is available at https://github.com/CQU-3DTEAM/TS-CATMA .