面向胎盘植入产前诊断的医学语义特征提取算法

发布时间：2018-04-14 11:36

本文选题：胎盘植入(PA) + 特征选择　；参考：《模式识别与人工智能》2015年06期

【摘要】：胎盘植入由于其临床特征隐匿,尚无一种敏感性、特异性高的产前诊断手段,因此文中将数据的特征提取方法引入胎盘植入产前诊断领域,从特征相关性的角度,提出胎盘植入有效医学语义的多目标特征优化问题,并给出求解该问题的一种改进的非支配排序遗传算法II(NSGA-II).基于实际胎盘植入相关临床数据的计算结果表明,文中算法能从复杂的胎盘植入相关临床数据中提取具有胎盘植入有效语义的特征集合.经过接收者操作特征(ROC)曲线分析,提取的特征医学语义具有较高的诊断价值,可为产科医师研究胎盘植入的发病机制和及时产前诊断提供有效的辅助手段.文中研究还发现,一些临床生化检查指标具有重要作用,可作为胎盘植入产前诊断的有效依据.
[Abstract]:Because the clinical features of placenta accreta are hidden, there is no sensitive and specific method for prenatal diagnosis. Therefore, the feature extraction method of data is introduced into the field of prenatal diagnosis of placenta accreta from the angle of feature correlation.A multi-objective feature optimization problem for effective medical semantics of placenta accreta is proposed, and an improved undominated sorting genetic algorithm (IIGA-IIA) is presented to solve the problem.The computational results based on the actual placenta accreta related clinical data show that the proposed algorithm can extract the effective feature set with placenta accreta effective semantics from the complicated placenta accreta related clinical data.The extracted feature medical semantics are of high diagnostic value through the analysis of receiver operating characteristics. It can provide an effective assistant means for obstetricians to study the pathogenesis of placenta accreta and timely prenatal diagnosis.It is also found that some clinical biochemical indicators play an important role in prenatal diagnosis of placenta accreta.
【作者单位】：福州大学数学与计算机科学学院;福建省妇幼保健院;
【基金】：国家卫生和计划生育委员会科研基金项目(No.WKJ-FJ-09)资助
【分类号】：R714.462;TP391.1

【相似文献】