Scholarlink

<p>Epilepsy analysis (EA) is vital for understanding seizures, improving diagnosis, and guiding treatment. However, the complexity of deep learning models and fragmented tools often limits accessibility for non-experts. We present epilepsy analysis visualizer (EAViz), a modular, AI-powered Python toolbox for automated EA. EAViz integrates advanced deep learning models built upon CNNs, LSTMs, ResNets, and attention mechanisms, enabling EEG- and video-based seizure analysis. GPU acceleration is employed to improve computational efficiency. The toolbox comprises five key modules: (1) epilepsy syndrome classification &amp; seizure detection (ESC&amp;SD), (2) artifact detection, (3) spike detection, (4) spike ripple detection, and (5) video-based seizure detection. Each module includes preconfigured models, automated preprocessing, and an intuitive graphical user interface (GUI) with guided interaction, making it accessible to clinicians without programming expertise. EAViz has been validated using clinical EEG data from the Children’s Hospital of Zhejiang University, achieving high performance (e.g., ESC&amp;SD achieved AUCs of 99.95% and 99.78% on respective tasks). By combining advanced AI with usability-centered design, EAViz improves diagnostic efficiency and expands access to intelligent epilepsy analysis in clinical practice.</p>

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

EAViz: a user-friendly deep learning-based epilepsy analysis visualizer using multimodal data

  • 期刊论文

  • 发表日期：

    2025年6月6日

查看全文
  • Ze Xia,
  • Dinghan Hu,
  • Tiejia Jiang,
  • Shuangpeng Zhu,
  • Xiaohui Lou,
  • Jiuwen Cao

摘要

Epilepsy analysis (EA) is vital for understanding seizures, improving diagnosis, and guiding treatment. However, the complexity of deep learning models and fragmented tools often limits accessibility for non-experts. We present epilepsy analysis visualizer (EAViz), a modular, AI-powered Python toolbox for automated EA. EAViz integrates advanced deep learning models built upon CNNs, LSTMs, ResNets, and attention mechanisms, enabling EEG- and video-based seizure analysis. GPU acceleration is employed to improve computational efficiency. The toolbox comprises five key modules: (1) epilepsy syndrome classification & seizure detection (ESC&SD), (2) artifact detection, (3) spike detection, (4) spike ripple detection, and (5) video-based seizure detection. Each module includes preconfigured models, automated preprocessing, and an intuitive graphical user interface (GUI) with guided interaction, making it accessible to clinicians without programming expertise. EAViz has been validated using clinical EEG data from the Children’s Hospital of Zhejiang University, achieving high performance (e.g., ESC&SD achieved AUCs of 99.95% and 99.78% on respective tasks). By combining advanced AI with usability-centered design, EAViz improves diagnostic efficiency and expands access to intelligent epilepsy analysis in clinical practice.