量子机器学习预测食管癌放疗前放射性食管炎
本研究纳入医院1的218例食管癌患者用于训练与内部验证、医院2的55例用于外部验证,采用混合量子-经典机器学习模型进行放疗前放射性食管炎(RE)预测。融合量子特征、剂量组学特征与临床因素后,Q-CNN+Q-attention模型外部验证准确率0.83、AUC 0.83,优于仅用量子特征模型。
要点
- 融合量子特征、剂量组学与临床因素可提升RE预测准确性。
- Q-CNN+Q-attention融合模型外部验证AUC为0.83。
- 仅量子特征模型外部验证准确率0.70–0.83。
核心结论
- 混合量子-经典机器学习方法可用于食管癌放疗前RE预测,多模态特征融合提升预测准确性与可靠性。
研究设计与样本
- 研究纳入医院1的218例食管癌患者进行训练与内部验证,医院2的55例进行外部验证,≥2级RE发生率分别为64例和20例。
- 剂量分布图像经角度编码转换为量子态,采用3种量子模型提取量子特征,并与手工剂量组学特征及临床因素进行特征级拼接,由随机森林分类器预测RE。
主要发现
- 仅用量子特征时,外部验证准确率分别为Q-CNN 0.70、Q-CNN+attention 0.83、Q-CNN+Q-attention 0.80。
- Q(Q-CNN+Q-attention)+D+C模型在训练、内部验证和外部验证的准确率分别为0.85、0.80、0.83,敏感度分别为0.83、0.73、0.73,特异度分别为0.92、0.84、0.89。
- Q(Q-CNN+Q-attention)+D+C模型的AUC在训练、内部验证和外部验证分别为0.89、0.89、0.83,优于仅用Q-CNN+Q-attention模型的0.78、0.78、0.80。
关键数值与推荐
- RE预测性能以准确率、敏感度、特异度和AUC评价,外部验证中最佳融合模型Q(Q-CNN+Q-attention)+D+C的AUC为0.83。
临床怎么用
- 该模型可在放疗前整合剂量分布图像、剂量组学特征与临床因素,辅助识别RE高风险患者。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
PURPOSE: To introduce a hybrid quantum-classical machine learning approach and validate its feasibility and accuracy for pretreatment radiation-induced esophagitis (RE) prediction in patients with esophageal cancer undergoing radiation therapy or chemoradiotherapy. METHODS AND MATERIALS: This study enrolled 218 patients with esophageal cancer from hospital 1 for training and internal validation and 55 patients with esophageal cancer from hospital 2 for external validation, with grade ≥2 RE incidences of 64 and 20, respectively. Dose distribution images were converted into quantum states via angle encoding. Quantum features (Qs) were extracted using 3 quantum models: (1) classical convolutional neural network (CNN) with quantum convolution (Q-CNN), (2) Q-CNN plus classical attention (Q-CNN + attention), and (3) Q-CNN plus quantum attention (Q-CNN + Q-attention). The hybrid machine learning model integrates handcrafted dosiomic features (Ds), Qs, and clinical factors (C) through a feature-level concatenation. The concatenated feature vector was processed by a Random Forest classifier for RE prediction. RESULTS: Models using only Qs achieved an accuracy of 0.70 (Q-CNN), 0.83 (Q-CNN + attention), and 0.80 (Q-CNN + Q-attention) in external validation, respectively. Feature fusion (Q + D + C) improved the accuracy of models in comparison with Qs alone. Q (Q-CNN + Q-attention) + D + C demonstrated optimal performance, achieving accuracy, sensitivity, and specificity of 0.85, 0.83, 0.92 (training); 0.80, 0.73, 0.84 (internal validation); and 0.83, 0.73, 0.89 (external validation), respectively. The Random Forest model using fused features Q + D + C extracted via Q-CNN + Q-attention achieved AUCs of 0.89 (training), 0.89 (internal validation), and 0.83 (external validation), outperforming models using only Q-CNN + Q-attention (AUCs: 0.78 training, 0.78 internal validation, and 0.80 external validation). CONCLUSIONS: This study proposes a novel quantum-classical machine learning method for pretreatment RE prediction. By integrating quantum amplitude encoding, quantum attention mechanisms, and multimodal feature fusion (Q + D + C), the model enhances prediction accuracy and reliability, demonstrating significant potential for clinical application.
引用
- Xie C, Shen Y, He J, et al. Pretreatment Radiation Esophagitis Prediction Using Quantum Machine Learning in Patients With Esophageal Cancer. Int J Radiat Oncol Biol Phys. 2026. PMID: 41850494. doi:10.1016/j.ijrobp.2026.03.010
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。