粗集理论及其在心电图自动分析中的应用研究

【作者】 张勇；

【导师】 舒兰；

【作者基本信息】 电子科技大学 ， 应用数学， 2000， 硕士

【摘要】 本文在粗糙集理论的基础上，对粗集决策表的简化理论进行了研究，并将粗集决策表简化方法用于心电图特征点的识别，得到了心电图特征点识别的简化规则，利用这些简化规则可以有效地识别出心电图特征点。本文还提出了一种确定特征元素类的方法—特征元确定算法，并将该方法用于心电图特征点的识别，提高了心电图特征点识别的准确度。本文还将决策表简化后得到的心电图特征点识别规则和特征元确定算法相结合，对决策表简化后的结果进行了验证和讨论。本文还研究了差分法原理，提出了k点差分中“k”的确定方法。并建立了心电图自动分析的算法，如R波识别、QRS波起止点识别、P波、T波识别算法，通过MIT-BIH的8个小时数据检验得到R波的检出率平均达95.65％。采用决策表简化后得到的心电图特征点识别规则对R波识别的检出率稍高于常规识别方法。本文的算法均采用MATLAB数学软件编写程序实现；同时为了便于心电图自动分析算法的研究与开发，还开发了一套用于心电图自动分析算法研究的MATLAB工具包，为心电图自动诊断部分奠定了基础。更多还原

【Abstract】 In this paper, Based on Rough sets, the algorithm of decision form reduction algorithm of Rough sets is studied, and is applied to the recognition of characteristic points in electrocardiogram. Finally the reduction rules are obtained. Moreover the confirmation way for characteristic elements is also proposed, and is applied to the recognition of electrocardiogram. Then the decision form reduction algorithm and confirmation of character element is combined, and applied to the recognition of characteristic points in electrocardiogram, and the two algorithms are validated and discussed. Finally the theory of difference is researched and a confirmation method of k in k-difference is given, and the algorithms of computer-aided Electrocardiogram about the recognition of QRS wave ~. P wave~. T wave are obtained. The algorithm for R wave is tested with 8-hour patient data of MIT-BIH. QRS, and average sensitivity of 95.65 percent is achieved. Then, the decision form algorithm for R wave is also tested, and average sensitivity of 96.39 percent is achieved. All algorithms are programmed under the development environment of MATLAB, which is a kind of mathematical software. For the convenience of research and development for electrocardiogram, some program toolbox for electrocardiogram are set up. This paper set up the basis for the diagnosis of electrocardiogram.更多还原