发表机构
Lebanese American University; Institute of Applied Artificial Intelligence, TÉLUQ University（黎巴嫩美国大学; 应用人工智能研究所，泰鲁克大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究针对基于脑电的癫痫检测面临的注释稀缺和类别不平衡问题，提出DiffEEG自监督基础模型，通过去噪扩散预训练和强化学习微调学习通用神经表示，在癫痫检测中取得较好效果，证明该方法能以最少标记数据实现临床可部署监测。
AI 中文摘要
基于脑电图（EEG）的癫痫检测深度学习面临严重挑战：注释稀缺和类别极度不平衡，发作事件在临床记录中占比不到10%。我们提出DiffEEG，一个有960万个参数的自监督基础模型，通过去噪扩散预训练和基于强化学习（RL）的微调解决这两个限制。在130万个来自Temple大学医院癫痫语料库（TUHSZ）的未标记片段上预训练，DiffEEG通过带多头自注意力的一维U-Net学习通用神经表示。对于下游适应，强化决策层采用策略梯度优化直接最大化F1分数，优先考虑对罕见癫痫事件的敏感性而非总体准确性。在严格的患者级评估（279名患者，留一折交叉验证）中，DiffEEG在4类癫痫亚型分类中达到61%的准确率和59%的F1分数，在二元检测中达到81%的准确率和85%的加权F1分数，尽管存在极端不平衡（患病率6.7%）仍保持临床上可行的癫痫召回率（59%）。片段级评估确定了97.6%准确率的上限，证实了强大的架构能力。DiffEEG表明，基于扩散的预训练与度量感知强化学习相结合，能够以最少的标记数据需求实现临床可部署的癫痫监测。
英文摘要
Deep learning for EEG-based seizure detection faces critical challenges: severe annotation scarcity and extreme class imbalance, where ictal events comprise less than 10\% of clinical recordings. We present DiffEEG, a 9.6M-parameter self-supervised foundation model that addresses both limitations through denoising diffusion pre-training and reinforcement learning (RL)-based fine-tuning. Pre-trained on 1.3M unlabeled segments from the Temple University Hospital Seizure Corpus (TUHSZ), DiffEEG learns generic neural representations via a 1D U-Net with multi-head self-attention. For downstream adaptation, a reinforced decision layer employs policy gradient optimization to directly maximize F1-score, prioritizing sensitivity to rare seizure events over overall accuracy. Under strict patient-wise evaluation (279 patients, Leave-One-Fold-Out), DiffEEG achieves 61\% accuracy and 59\% F1 for 4-class seizure subtyping, and 81\% accuracy with 85\% weighted F1 for binary detection, maintaining clinically viable seizure recall (59\%) despite extreme imbalance (6.7\% prevalence). Segment-level evaluation establishes an upper bound of 97.6\% accuracy, confirming strong architectural capacity. DiffEEG demonstrates that diffusion-based pre-training combined with metric-aware reinforcement learning enables clinically deployable seizure monitoring with minimal labeled data requirements.
Comments19 pages, 6 figures
