基于深度主动学习的智能胎心监护模型研究  

Research on intelligent fetal heart monitoring model based on deep active learning

作　　者：全斌 黄雅静 李艳芳[3] 陈沁群[1,4] 张洪来[1,4,5] 李丽 刘桂清 魏航[1,4] QUAN Bin;HUANG Yajing;LI Yanfang;CHEN Qinqun;ZHANG Honglai;LI Li;LIU Guiqing;WEI Hang(School of Medical Information Engineering,Guangzhou University of Chinese Medicine,Guangzhou 510006,P.R.China;National Engineering Laboratory of Big Data System Computing Technology,Shenzhen University,Shenzhen,Guangdong 518060,P.R.China;First Affiliated Hospital of Guangzhou University of Chinese Medicine,Guangzhou 510405,P.R.China;Intelligent Chinese Medicine Research Institute,Guangzhou University of Chinese Medicine,Guangzhou 510006,P.R.China;State Key Laboratory of Traditional Chinese Medicine Syndrome,Guangzhou University of Chinese Medicine,Guangzhou 510006,P.R.China;Guangzhou Sunray Medical Apparatus Co.Ltd,Guangzhou 510620,P.R.China;Tianhe District People's Hospital,First Affiliated Hospital of Jinan University,Guangzhou 510630,P.R.China)

机构地区：[1]广州中医药大学医学信息工程学院,广州510006 [2]深圳大学大数据系统计算技术国家工程实验室,广东深圳518060 [3]广州中医药大学第一附属医院,广州510405 [4]广州中医药大学智能中医研究院,广州510006 [5]广州中医药大学中医症候全国重点实验室,广州510006 [6]广州三瑞医疗器械有限公司,广州510620 [7]暨南大学附属第一医院天河区人民医院,广州510630

出　　处：《生物医学工程学杂志》2025年第1期57-64,共8页Journal of Biomedical Engineering

基　　金：国家自然科学基金项目(71804031);广东省医学科研基金面上项目(A2019428)。

摘　　要：胎心宫缩监护图(CTG)是一种无侵入式的产前胎儿窘迫诊断的重要工具。面向基于深度学习的智能胎心监护需求,本文提出了基于三支决策理论(TWD)和多目标优化主动学习(MOAL)的深度主动学习算法(TWD-MOAL)。在卷积神经网络(CNN)分类模型的训练周期过程中,该算法结合TWD理论,在粒度批处理模式下选择高置信度样本作为伪标记样本,同时参考由产科专家标注的低置信度样本,在课题组整理的16355条产前CTG信号数据集上进行验证。实验结果显示,本文所提算法仅利用40%的已标注样本,就已经达到80.63%的准确率;从各项指标综合来看,都优于其他框架下的主动学习算法。研究表明,本文所提出的基于TWDMOAL的智能胎心监护模型合理可行,明显减少产科专家标注样本所耗费的时间和精力,有效解决了临床CTG信号数据不平衡问题,这对于辅助临床医师判读以及实现胎心监护模型的智能化具有重要意义。Cardiotocography(CTG)is a non-invasive and important tool for diagnosing fetal distress during pregnancy.To meet the needs of intelligent fetal heart monitoring based on deep learning,this paper proposes a TWDMOAL deep active learning algorithm based on the three-way decision(TWD)theory and multi-objective optimization Active Learning(MOAL).During the training process of a convolutional neural network(CNN)classification model,the algorithm incorporates the TWD theory to select high-confidence samples as pseudo-labeled samples in a fine-grained batch processing mode,meanwhile low-confidence samples annotated by obstetrics experts were also considered.The TWD-MOAL algorithm proposed in this paper was validated on a dataset of 16355 prenatal CTG records collected by our group.Experimental results showed that the algorithm proposed in this paper achieved an accuracy of 80.63%using only 40%of the labeled samples,and in terms of various indicators,it performed better than the existing active learning algorithms under other frameworks.The study has shown that the intelligent fetal heart monitoring model based on TWD-MOAL proposed in this paper is reasonable and feasible.The algorithm significantly reduces the time and cost of labeling by obstetric experts and effectively solves the problem of data imbalance in CTG signal data in clinic,which is of great significance for assisting obstetrician in interpretations CTG signals and realizing intelligence fetal monitoring.

关 键 词：智能胎心监护 深度主动学习 三支决策 多目标优化 卷积神经网络 

分 类 号：TN911.7[电子电信—通信与信息系统] TP18[电子电信—信息与通信工程] R714.5[自动化与计算机技术—控制理论与控制工程]