南方医科大学学报

南方医科大学学报 ›› 2026, Vol. 46 ›› Issue (7): 1714-1722.doi: 10.12122/j.issn.1673-4254.2026.07.24

• • 上一篇    

医疗先验引导的TabMap深度学习模型用于卵巢癌预测及可解释性研究

朱君1(), 谭顺谦1, 黄方俊2, 蔡光瑶3, 甄鑫1()   

  1. 1.南方医科大学生物医学工程学院，广东 广州 510515
    2.南方医科大学附属广东省人民医院（广东省医学科学院）肿瘤放射治疗科，广东 广州 510080
    3.中山大学肿瘤防治中心妇科//华南肿瘤学国家重点实验室//肿瘤医学省部共建协同创新中心，广东 广州 510060
  • 收稿日期:2025-12-03 出版日期:2026-07-20 发布日期:2026-07-20
  • 通讯作者: 甄鑫 E-mail:2570540683@qq.com;xinzhen@smu.edu.cn
  • 作者简介:朱 君，在读硕士研究生，E-mail:2570540683@qq.com
  • 基金资助:
    国家自然科学基金(82572381);国家自然科学基金(82404078);广东省基础与应用基础研究基金(2024A1515012100);广东省基础与应用基础研究基金项目区域联合基金-青年基金项目(2023A1515110701)

Medical prior-guided TabMap deep learning model for ovarian cancer prediction and interpretability analysis

Jun ZHU1(), Shunqian TAN1, Fangjun HUANG2, Guangyao CAI3, Xin ZHEN1()   

  1. 1.School of Biomedical Engineering, Southern Medical University, Guangzhou 510515, China
    2.Department of Radiation Oncology, Guangdong Provincial People's Hospital (Guangdong Academy of Medical Sciences), Southern Medical University, Guangzhou 510080, China
    3.Department of Gynecology, Sun Yat-sen University Cancer Center, South China State Key Laboratory of Oncology, Provincial-Ministry Collaborative Innovation Center for Medical Oncology, Guangzhou 510060, China
  • Received:2025-12-03 Online:2026-07-20 Published:2026-07-20
  • Contact: Xin ZHEN E-mail:2570540683@qq.com;xinzhen@smu.edu.cn
  • Supported by:
    National Natural Science Foundation of China(82572381)

RichHTML

9

PDF

97

摘要：

目的 针对卵巢癌早期诊断中表格型医疗数据特征关联复杂、模型可解释性不足等问题，提出一种融合医疗先验知识的TabMap图像映射与深度学习预测框架。 方法 首先，根据临床语义将36个医疗诊断特征划分为血常规分区、生化指标分区、肿瘤标志物分区以及其他（凝血、炎症及临床变量）分区4个空间连续区域；其次，采用Gromov-Wasserstein最优传输算法求解特征空间与像素空间的最优耦合，生成保持拓扑结构的TabMap图像；然后，设计融合SE注意力机制和全局平均池化的轻量级卷积神经网络，通过类别加权损失和加权采样策略应对样本不平衡问题；最后，利用类激活映射技术可视化模型决策过程，实现可解释性分析。 结果 在真实卵巢癌数据集上的实验表明，本文方法的测试集准确率达到91.82%，精确率为89.96%，召回率为89.45%，F1分数为0.8970，平衡准确率为88.51%，相比传统机器学习方法准确率提升8.26%~14.93%，相比标准深度学习模型准确率提升2.28%~13.60%。可解释性分析显示，模型在肿瘤标志物分区以及凝血、炎症以及临床变量相关分区中呈现出明显的响应模式，与临床既有认知保持一致；特征重要性排序进一步强调了CA125、HE4等关键指标在模型决策中的核心作用。 结论 本研究提出的医疗先验引导TabMap方法有效融合了领域知识与数据驱动学习，在提升预测性能的同时增强了模型的临床可解释性，为表格型医疗数据的深度学习建模提供了新思路，具有良好的临床应用潜力。

关键词: 卵巢癌预测, TabMap, 深度学习, 可解释性, 医疗先验知识

Abstract:

Objective To develop a TabMap image mapping and deep learning prediction framework that integrates medical prior knowledge to address the challenges of complex feature associations in tabular medical data and insufficient model interpretability in early ovarian cancer diagnosis. Methods Based on clinical semantics, 36 medical diagnostic features were partitioned into 4 spatially continuous regions, namely the routine blood test partition, biochemical indicators partition, tumor markers partition, and other (coagulation, inflammation, and other clinical variables) partition. The Gromov-Wasserstein optimal transport algorithm was then employed to solve the optimal coupling between feature space and pixel space, thus generating TabMap images that preserve the topological structures. A lightweight convolutional neural network incorporating SE attention mechanism and global average pooling was designed to handle sample imbalance using class-weighted loss and weighted sampling strategies. Finally, class activation mapping (CAM) technique was utilized to visualize the model's decision-making process for interpretability analysis. Results Experiments on a real ovarian cancer dataset demonstrated that the proposed method achieved a test accuracy of 91.82% with a precision of 89.96%, recall of 89.45%, F1-score of 0.8970 and balanced accuracy of 88.51%, representing accuracy improvements of 8.26%-14.93% over traditional machine learning methods and 2.28%-13.60% over standard deep learning models. Interpretability analysis showed that the model exhibits pronounced activation patterns in the tumor-marker region as well as in coagulation-, inflammation-, and age-related zones, consistent with established clinical understanding. The feature-importance ranking further underscored the pivotal roles of CA125, HE4, and other key biomarkers in guiding the model's decisions. Conclusion The proposed medical prior-guided TabMap method effectively integrates domain knowledge with data-driven learning, which enhances both its prediction performance and clinical interpretability. This strategy provides a novel approach for deep learning modeling of tabular medical data with good clinical potentials.

Key words: ovarian cancer prediction, TabMap, interpretability, medical prior knowledge, deep learning