基于18F-FDG PET/CT的全自动三维多模态深度学习模型用于结直肠癌术前T分期预测
回顾性纳入474例结直肠癌患者(内部队列400例、外部队列74例),构建基于TotalSegmentator自动分割和3D ResNet18结合交叉多头注意力机制的全自动多模态模型预测早期(T1-T2)与进展期(T3-T4)T分期。CT+PET+临床模型与CT+PET模型在内外部验证中AUC均优于单模态模型和放射科医师,Grad-CAM显示早期T分期关注肿瘤区域、进展期关注瘤周组织浸润。
要点
- CT+PET+临床与CT+PET模型内部验证AUC均为0.869,优于PET、CT、临床单模态模型
- 外部验证AUC分别为0.814、0.812,单模态模型为0.763、0.714、0.663
- 两模型AUC均高于放射科医师(AUC 0.627),除临床模型外差异有统计学意义(P<0.05)
核心结论
- 全自动多模态模型融合PET/CT与交叉多头注意力机制,改善了结直肠癌T分期预测,优于单模态模型和放射科医师,提供省时的辅助临床决策工具。
研究设计与样本
- 回顾性队列纳入474例结直肠癌患者,内部队列400例,外部队列74例。
- 患者分为早期T分期(T1-T2)组和进展期T分期(T3-T4)组。
- 基于TotalSegmentator实现感兴趣体积(VOI)自动分割,构建基于3D ResNet18并结合交叉多头注意力机制的深度学习模型。
- 比较五种模型(CT+PET+临床(CPC)、CT+PET(CP)、PET(P)、CT(C)、临床)与两名放射科医师的评估。
- 采用受试者工作特征曲线下面积(AUC)评价性能,并用Grad-CAM提供决策关键区域的可视化解释。
主要发现
- 自动分割的Dice评分在CT为0.884、PET为0.888。
- 内部验证队列中CPC和CP模型AUC均为0.869,优于单模态模型(PET 0.832、CT 0.809、临床 0.728)和放射科医师(AUC 0.627);除临床模型外,所有模型与放射科医师比较P<0.05。
- 外部验证呈现相似趋势,CPC、CP、PET、CT、临床模型及放射科医师的AUC分别为0.814、0.812、0.763、0.714、0.663和0.704。
- Grad-CAM可视化显示早期T分期以肿瘤为中心区域为主,进展期T分期以瘤周组织浸润为主。
关键数值与推荐
- 自动分割Dice评分:CT 0.884、PET 0.888。
- 内部验证AUC:CPC 0.869、CP 0.869、PET 0.832、CT 0.809、临床 0.728、放射科医师 0.627。
- 外部验证AUC:CPC 0.814、CP 0.812、PET 0.763、CT 0.714、临床 0.663、放射科医师 0.704。
- 除临床模型外,所有模型与放射科医师比较P<0.05。
临床怎么用
- 原文未据此给出推荐。
局限性
- 原文未报告局限性。
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: This study aimed to develop a fully automated 3D multi-modal deep learning model using preoperative 18F-FDG PET/CT to predict the T-stage of colorectal cancer (CRC) and evaluate its clinical utility. METHODS: A retrospective cohort of 474 CRC patients was included, with 400 patients for internal cohort and 74 patients for external cohort. Patients were classified into early T-stage (T1-T2) and advanced T-stage (T3-T4) groups. Automatic segmentation of the volume of interest (VOI) was achieved based on TotalSegmentator. A 3D ResNet18-based deep learning model integrated with a cross-multi-head attention mechanism was developed. Five models (CT + PET + Clinic (CPC), CT + PET (CP), PET (P), CT (C), Clinic) and two radiologists' assessment were compared. Performance was evaluated using Area Under the Curve (AUC). Grad-CAM was employed to provide visual interpretability of decision-critical regions. RESULTS: The automated segmentation achieved Dice scores of 0.884 (CT) and 0.888 (PET). The CPC and CP models achieved superior performance, with AUCs of 0.869 and 0.869 in the internal validation cohort, respectively, outperforming single-modality models (P: 0.832; C: 0.809; Clinic: 0.728) and the radiologists (AUC: 0.627, P < 0.05 for all models vs. radiologists, except for the Clinical model). External validation exhibited a similar trend, with AUCs of 0.814, 0.812, 0.763, 0.714, 0.663 and 0.704, respectively. Grad-CAM visualization highlighted tumor-centric regions for early T-stage and peri-tumoral tissue infiltration for advanced T-stage. CONCLUSION: The fully automated multimodal, fusing PET/CT with cross-multi-head-attention, improved T-stage prediction in CRC, surpassing the single-modality models and radiologists, offering a time-efficient tool to aid clinical decision-making.
引用
- Zhang M, Li Y, Zheng C, et al. Fully automated 3D multi-modal deep learning model for preoperative T-stage prediction of colorectal cancer using 18F-FDG PET/CT. Eur J Nucl Med Mol Imaging. 2025. PMID: 40719866. doi:10.1007/s00259-025-07450-5
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。