首次系统评估13种可解释AI方法，为头颈癌预后预测提供可信工具。
Ranking XAI Methods for Head and Neck Cancer Outcome Prediction
- 用24项指标全面对比13种XAI方法，覆盖准确性与可靠性
- 集成梯度和DeepLIFT在可解释性上表现最优，稳定领先
- 研究结果可推广至其他医学影像任务，助力临床应用
头颈癌（HNC）患者预后预测有助于制定个性化治疗方案。利用先进人工智能技术处理PET/CT数据，已广泛用于提升HNC预后预测性能，但AI模型的可解释性仍是其临床应用的关键障碍。与以往依赖经验选择可解释AI（XAI）方法的研究不同，本工作首次在多中心HECKTOR挑战数据集上，对13种XAI方法进行跨24项指标的综合评估与排名，涵盖忠实性、鲁棒性、复杂度与合理性。实验结果显示，不同方法在各项评价维度上差异显著，其中集成梯度（IG）和DeepLIFT（DL）在忠实性、复杂度和合理性方面均持续获得高分。该研究强调了全面评估XAI方法的重要性，并可拓展至其他医学影像任务。
原文摘要 · Abstract (English)
For head and neck cancer (HNC) patients, prognostic outcome prediction can support personalized treatment strategy selection. Improving prediction performance of HNC outcomes has been extensively explored by using advanced artificial intelligence (AI) techniques on PET/CT data. However, the interpretability of AI remains a critical obstacle for its clinical adoption. Unlike previous HNC studies that empirically selected explainable AI (XAI) techniques, we are the first to comprehensively evaluate and rank 13 XAI methods across 24 metrics, covering faithfulness, robustness, complexity and plausibility. Experimental results on the multi-center HECKTOR challenge dataset show large variations across evaluation aspects among different XAI methods, with Integrated Gradients (IG) and DeepLIFT (DL) consistently obtained high rankings for faithfulness, complexity and plausibility. This work highlights the importance of comprehensive XAI method evaluation and can be extended to other medical imaging tasks.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。