宫缩曲线分析及其状态实时识别算法的研究

发布时间：2018-04-26 16:25

  本文选题：宫缩状态 + 基线估计　； 参考：《生物医学工程学杂志》2017年05期

【摘要】：宫缩状态实时识别在分娩镇痛中具有重要意义,但相关传统算法和系统无法满足实时识别宫缩状态的要求。针对上述情况,本文设计了一套宫缩状态实时分析算法。该算法包括宫缩信号预处理、基于直方图和线性迭代的宫缩基线估计以及一种基于有限状态机原理的实时识别算法,可根据前一点的宫缩状态以及一系列状态转换条件来识别当前的宫缩状态,并且设置缓冲机制来避免不真实的状态转换。为了评估该算法的性能表现,本文将其与现有的一种电子胎儿监护仪的宫缩分析算法进行比较。实验结果表明,本文算法能够在宫缩信号实时监测的同时对宫缩状态进行实时分析,算法敏感度为0.939 9,阳性预测值为0.869 3,具有较高的准确度,可达到临床监测的要求。
[Abstract]:The real-time recognition of uterine contractions is of great significance in labor analgesia, but the traditional algorithms and systems can not meet the requirements of real-time recognition of uterine contractions. In view of the above situation, this paper designs a set of real-time analysis algorithm of uterine contraction state. The algorithm includes preprocessing of uterine contraction signal, histogram and linear iterative baseline estimation of uterine contraction, and a real-time recognition algorithm based on the principle of finite state machine. The current state of uterine contraction can be identified according to the contractive state of the former point and a series of state transition conditions, and a buffer mechanism is set to avoid the untrue state transition. In order to evaluate the performance of the algorithm, this paper compares it with an existing algorithm for uterine contraction analysis of an electronic fetal monitor. The experimental results show that the algorithm can analyze the state of uterine contraction at the same time as the real-time monitoring of uterine contraction signal. The sensitivity of the algorithm is 0.939 9 and the positive predictive value is 0.869 3. The algorithm has a high accuracy and can meet the requirements of clinical monitoring.
【作者单位】： 暨南大学信息科学技术学院电子工程系;
【基金】：国家国际科技合作专项资助项目(2015DFI12970) 粤港共性技术招标资助项目(2013B010136002) 广东省科技计划应用型科技研发专项资助项目(2015B020233010) 广东省科技计划重点资助项目(2015B020214004)
【分类号】：R714;TN911.7
，

本文编号：1806773