融入图像的心肺复苏评价算法研究
发布时间：2018-08-21 07:29
【摘要】：心肺复苏已成为全球最为推崇且普及最为广泛的急救技术，其技能培训对于急救护理学教学有着重要意义。随着数字医疗技术、计算机科学技术的发展，涌现了大量用于心肺复苏教学训练的设备，其中最具有代表性的是医疗培训模拟人。但是，目前的心肺复苏模拟人教学系统，仅从模拟病人内部体征的感知角度，采集了操作者的按压深度、呼吸等数据，进而对操作者的动作做出判别，而忽略了操作者的按压姿势、手形、垂直用力的正确性，可见系统还不够全面，需要进一步的完善。同时，在心肺复苏操作效果评价方面，心肺复苏操作效果与操作者的按压深度、按压位置、按压频率以及呼吸量、呼吸频率、垂直用力、正确的姿势、手形等诸多因素有关，如何将这些影响因素通过一定的方法进行分析、整合，形成对心肺复苏操作的综合评价，是目前心肺复苏教学领域需要解决的问题之一。因此，本文针对这几方面展开研究，具体工作如下： 心肺复苏教学评价模型的总体设计，包括利用模拟人内部传感器采集按压、呼吸数据，利用图像传感器识别按压姿势、垂直用力、手形，以及综合各传感器的数据得出综合评价结果等三个部分。其中，心肺复苏手形识别以及综合评价模型的建立是本文的重点和难点。 在心肺复苏手形识别方面，本文借鉴静态手势识别方法，设计了一种基于组合特征的心肺复苏手形识别方法。首先，通过椭圆肤色模型以及对图像进行相应预处理，，获得按压手形二值图像；在特征提取中，本文提出了一种基于轮廓凸包和凹陷的结构特征提取算法，利用手指个数、手指指尖夹角关系等手形结构特征作为局部特征，并利用改进的傅里叶描述子作为全局特征，形成心肺复苏手形的组合特征；在手形识别中，根据局部特征和全局特征的各自特点，设计了一种逐步排除的快速识别方法，最后利用基于欧氏距离的模版匹配方法进行识别。实验结果表明，本文的方法可以有效的区分正确、错误按压手形。 在心肺复苏评价方面，为了综合各传感器采集的评价指标数据，对学生的心肺复苏操作做出客观全面的综合评价，本文引入数据融合的思想，提出基于支持向量回归的决策级评价模型。本文选用混合核函数构造支持向量回归模型，并利用混沌差分进化算法对混合核SVR的参数进行优化选择，进一步提高了模型的拟合精度和泛化能力，然后利用改进的支持向量回归模型对学生的心肺复苏操作进行综合评价。 最后，通过实验验证本文所提方法的有效性。实验表明，本文的评价模型可以有效的融合传感器采集的各项指标因素，得出综合评价结果，从而为心肺复苏教学系统建立了科学的评价体系。
[Abstract]:Cardiopulmonary resuscitation (CPR) has become the most popular and widely used first aid technology in the world. With the development of digital medical technology and computer science and technology, a large number of equipment used in cardiopulmonary resuscitation teaching and training have emerged, among which the most representative is the medical training simulator. However, the current teaching system for simulating cardiopulmonary resuscitation (CPR) only collects the data of the operator's pressing depth, breathing and so on from the perspective of the perception of the internal physical signs of the simulated patient, and then makes a judgment on the action of the operator. The correctness of the operator's pressing posture, hand shape and vertical force is ignored, so the system is not comprehensive enough and needs further improvement. At the same time, in the evaluation of the operational effect of cardiopulmonary resuscitation, the operational effect of CPR is related to the operator's pressing depth, pressing position, pressing frequency, breathing quantity, respiratory frequency, vertical force, correct posture, hand shape and so on. It is one of the problems that need to be solved in the teaching field of cardiopulmonary resuscitation (CPR) how to analyze and integrate these influencing factors through certain methods to form a comprehensive evaluation of the operation of cardiopulmonary resuscitation (CPR). Therefore, this paper studies these aspects, the specific work is as follows: the overall design of teaching evaluation model of cardiopulmonary resuscitation, including the use of simulated human internal sensors to collect compression, respiratory data, The image sensor is used to identify the pressing position, the vertical force, the hand shape, and the synthetic evaluation result by synthesizing the data of each sensor. Among them, the recognition of hand shape and the establishment of comprehensive evaluation model are the key and difficult points in this paper. In the aspect of hand shape recognition of CPR, a hand recognition method based on combined features is designed by using static gesture recognition method. Firstly, by using the elliptical skin model and the corresponding preprocessing of the image, the binary image of the pressing hand is obtained. In the feature extraction, a structural feature extraction algorithm based on the contour convex hull and the depression is proposed, using the number of fingers, the number of fingers, The finger fingertip angle relation is used as the local feature, and the improved Fourier descriptor is used as the global feature to form the combined character of the hand shape of cardiopulmonary resuscitation. According to the respective characteristics of local and global features, a fast recognition method with gradual exclusion is designed. Finally, the template matching method based on Euclidean distance is used to identify. The experimental results show that the proposed method can effectively distinguish the correct and wrong pressing hand shape. In the evaluation of cardiopulmonary resuscitation (CPR), in order to synthesize the evaluation index data collected by various sensors and make an objective and comprehensive evaluation of the students' CPR operation, this paper introduces the idea of data fusion. A decision level evaluation model based on support vector regression is proposed. In this paper, the hybrid kernel function is used to construct the support vector regression model, and the chaotic differential evolution algorithm is used to optimize the parameters of the hybrid kernel SVR, which further improves the fitting accuracy and generalization ability of the model. Then the improved support vector regression model was used to evaluate the cardiopulmonary resuscitation (CPR). Finally, the effectiveness of the proposed method is verified by experiments. The experimental results show that the evaluation model of this paper can effectively fuse the various index factors collected by sensors and obtain the comprehensive evaluation results, thus establishing a scientific evaluation system for the teaching system of cardiopulmonary resuscitation.
【学位授予单位】：辽宁大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP391.41;R459.7
本文编号：2194994
[Abstract]:Cardiopulmonary resuscitation (CPR) has become the most popular and widely used first aid technology in the world. With the development of digital medical technology and computer science and technology, a large number of equipment used in cardiopulmonary resuscitation teaching and training have emerged, among which the most representative is the medical training simulator. However, the current teaching system for simulating cardiopulmonary resuscitation (CPR) only collects the data of the operator's pressing depth, breathing and so on from the perspective of the perception of the internal physical signs of the simulated patient, and then makes a judgment on the action of the operator. The correctness of the operator's pressing posture, hand shape and vertical force is ignored, so the system is not comprehensive enough and needs further improvement. At the same time, in the evaluation of the operational effect of cardiopulmonary resuscitation, the operational effect of CPR is related to the operator's pressing depth, pressing position, pressing frequency, breathing quantity, respiratory frequency, vertical force, correct posture, hand shape and so on. It is one of the problems that need to be solved in the teaching field of cardiopulmonary resuscitation (CPR) how to analyze and integrate these influencing factors through certain methods to form a comprehensive evaluation of the operation of cardiopulmonary resuscitation (CPR). Therefore, this paper studies these aspects, the specific work is as follows: the overall design of teaching evaluation model of cardiopulmonary resuscitation, including the use of simulated human internal sensors to collect compression, respiratory data, The image sensor is used to identify the pressing position, the vertical force, the hand shape, and the synthetic evaluation result by synthesizing the data of each sensor. Among them, the recognition of hand shape and the establishment of comprehensive evaluation model are the key and difficult points in this paper. In the aspect of hand shape recognition of CPR, a hand recognition method based on combined features is designed by using static gesture recognition method. Firstly, by using the elliptical skin model and the corresponding preprocessing of the image, the binary image of the pressing hand is obtained. In the feature extraction, a structural feature extraction algorithm based on the contour convex hull and the depression is proposed, using the number of fingers, the number of fingers, The finger fingertip angle relation is used as the local feature, and the improved Fourier descriptor is used as the global feature to form the combined character of the hand shape of cardiopulmonary resuscitation. According to the respective characteristics of local and global features, a fast recognition method with gradual exclusion is designed. Finally, the template matching method based on Euclidean distance is used to identify. The experimental results show that the proposed method can effectively distinguish the correct and wrong pressing hand shape. In the evaluation of cardiopulmonary resuscitation (CPR), in order to synthesize the evaluation index data collected by various sensors and make an objective and comprehensive evaluation of the students' CPR operation, this paper introduces the idea of data fusion. A decision level evaluation model based on support vector regression is proposed. In this paper, the hybrid kernel function is used to construct the support vector regression model, and the chaotic differential evolution algorithm is used to optimize the parameters of the hybrid kernel SVR, which further improves the fitting accuracy and generalization ability of the model. Then the improved support vector regression model was used to evaluate the cardiopulmonary resuscitation (CPR). Finally, the effectiveness of the proposed method is verified by experiments. The experimental results show that the evaluation model of this paper can effectively fuse the various index factors collected by sensors and obtain the comprehensive evaluation results, thus establishing a scientific evaluation system for the teaching system of cardiopulmonary resuscitation.
【学位授予单位】：辽宁大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP391.41;R459.7
【参考文献】
相关期刊论文 前1条
1 王雪松;程玉虎;郝名林;;一种支持向量机参数选择的改进分布估计算法[J];山东大学学报(工学版);2009年03期
本文编号：2194994
本文链接：https://www.wllwen.com/yixuelunwen/jjyx/2194994.html
最近更新
教材专著
