MR-HOCM患者连续多普勒频谱图的智能分析

发布时间：2018-06-02 00:22

本文选题：CWDS + MR-HOCM　；参考：《西华大学》2017年硕士论文

【摘要】：1958年,Teare首次全面描述了肥厚型心肌病(Hypertrophic Cardiomyopathy,HCM),临床中,根据静息状态下左室流出道压力阶差(Left Ventricle Outflow Tract Gradient,LVO TG)值将HCM分为肥厚梗阻型()和肥厚非梗阻型()。当肥厚梗阻型心肌病(Hypertrophic Obstructive Cardiomyopathy,HOCM)患者混合其他病变如二尖瓣返流(Mitral Regurgitation,MR)时,初级医师易过高估计LVOT的最大瞬时峰值流速和LVOTG值,影响病情判断和治疗决策的最优化。因此,在本研究室和第四军医大学第一附属医院西京医院HCM研究小组前期研究工作成果的基础上,本文致力于伴二尖瓣返流肥厚梗阻型心肌病(Hypertrophic Obstructive Cardiomyopathy with Mitral Regurgitation,MR-HOCM)患者LVOT连续多普勒频谱图(Continuous Wave Doppler Spectrum,CWDS-LVO T)的智能辅助测量,鉴别MR-HOCM患者的LVOT血流流速和LVO TG值时是否发生过高估计并对LVOTG值进行矫正,同时自动提取连续多普勒频谱图(Continuous Wave Doppler Spectrum,CWDS)的特征参数,为LVOT血流动力情况提供更加丰富的临床信息。为了完成以上目标,本文主要从以下几个方面展开深入的研究:(1)MR-HOCM患者CWDS-LVOT的采集和预处理深入研究MR-HOCM患者CWDS-LVOT的特点、采集方法及参数含义、LVOTG测量方法、准确度和误差原因;在第四军医大学第一附属医院西京医院住院一部超声科由刚入科不超过三年的初级超声诊断医师和具有超过二十年以上丰富经验的专家超声诊断医师采集12名MR-HOCM患者CWDS-LVOT,并对LVOTG值的测量结果进行确认和记录,并设计合理的降噪预处理方法,通过选取中值滤波,均值滤波和高斯曲率滤波三种图像滤波方法进行滤波效果的评估和对比,通过对比试验,高斯曲率滤波(Gaussian Curvature Filter,GCF)在MR-HOCM患者CWDS-LVOT的降噪预处理上,效果明显。(2)盲源分离(Blind Source Separation,BSS)算法及应用深入研究BSS实现的基本模型、经典算法FAST-ICA的数学理论和基本思想,研究算法的实现过程和相应的实现步骤;探究BSS过程中出现的盲不确定性出现的原因;根据MR-HOCM患者CWDS-LVOT特点,完成基于FAST-ICA的MR-HOCM患者CWDS-LVOT的信号分离算法和实验验证,同时使用占格率对估计源信号的之间的独立性进行了测度。将LVOTG估计值与专家诊断结果进行对比分析,使用灵敏度和特异性指标,完成了算法准确率的评估,并建立了基于Subspace discriminant集成分类器的算法评估模型。(3)MR-HOCM患者CWDS-LVOT特征参数的提取深入研究MR-HOCM患者CWDS-LVOT特征参数提取方法;利用单自由度模型提取MR-HOCM患者CWDS-LVOT的最大频率曲线,基于提取得到的最大频率曲线进行8个特征参数的提取,提取的特征参数包括收缩期最大流速S、舒张末期最低流速D、收缩舒张流速比SD、LVOTG值、阻力指数RI、搏动指数PI、收缩谱宽度W和收缩上升时间T,并利用得到的特征参数完成对血流状况的评估。(4)系统界面设计采用MATLAB GUI设计一个MR-HOCM患者CWDS-LVOT的简易分析鉴别系统。操作界面分LVOTG过高估计判定、特征参数提取和诊断分析报告三大部分。LVOTG过高估计判定部分,显示原始MR-HOCM患者CWDS-LVOT、估计源信号图、流速矫正后估计源信号和MR-HOCM患者CWDS-LVOT最大频率曲线的对比图以及矫正后流速和LVOTG值。特征参数提取和诊断分析报告部分,编写相应控件的回调函数,结合相应所需的临床诊断参数和患者信息,将特征参数和分析结果图生成WORD形式的分析报告。本文主要对MR-HOCM患者CWDS-LVOT进行了深入的研究分析,建立了一个MR-HOCM患者CWDS-LVOT的流速简易分析鉴别系统,实现了对MR-HOCM患者CWDS-LVOT是否产生流速和LVOTG值过高估计的判定和矫正,同时还结合临床所需,实现了MR-HOCM患者CWDS-LVOT部分特征参数的自动提取,并通过实验进一步验证了该系统的有效性和可行性。
[Abstract]:In 1958, Teare described the Hypertrophic Cardiomyopathy (HCM) for the first time. In clinical, the LVO pressure order (Left Ventricle Outflow Tract Gradient, LVO TG) is divided into hypertrophic obstructive () and hypertrophic non obstructive (Left). Cardiomyopathy, HOCM), when patients are mixed with other diseases such as Mitral Regurgitation (MR), the primary physician is prone to overestimate the maximum instantaneous peak velocity of LVOT and the LVOTG value, and affect the optimization of the condition judgment and treatment decision. Therefore, in this research room and the HCM research team of the Xijing Hospital of the First Affiliated Hospital of The Fourth Military Medical University On the basis of the results of the study, this paper aims to identify the blood flow of Hypertrophic Obstructive Cardiomyopathy with Mitral Regurgitation (MR-HOCM) patients with LVOT continuous Doppler spectrum (Continuous Wave Doppler Spectrum). Whether the flow rate and the LVO TG value are overestimated and correct the LVOTG value, and automatically extract the characteristic parameters of the continuous Doppler spectrum (Continuous Wave Doppler Spectrum, CWDS), provide more abundant clinical information for the LVOT blood flow power situation. Study: (1) the acquisition and preprocessing of CWDS-LVOT in MR-HOCM patients were deeply studied for the characteristics of CWDS-LVOT in MR-HOCM patients, the means of acquisition and parameters, the method of LVOTG measurement, the accuracy and the cause of error; a primary ultrasonic diagnostics in the Department of ultrasound of the First Affiliated Hospital of The Fourth Military Medical University, which was not more than three years in the Department of the Department of education. And expert ultrasonic diagnostics who have more than twenty years of experience to collect 12 MR-HOCM patients CWDS-LVOT, confirm and record the measurement results of LVOTG values, and design reasonable noise reduction preprocessing methods, and select the median filtering, mean filtering and Gauss curvature filtering for the filtering effect of three kinds of image filtering methods. Evaluation and comparison, the effect of Gaussian Curvature Filter (Filter, GCF) on CWDS-LVOT noise reduction preprocessing of MR-HOCM patients is obvious. (2) blind source separation (Blind Source Separation, BSS) algorithm and its application to the basic model of BSS implementation, the mathematical theory and basic idea of the classical algorithm FAST-ICA. The realization process of the algorithm and the corresponding implementation steps; explore the cause of the occurrence of the blind uncertainty in the BSS process. According to the CWDS-LVOT characteristics of the MR-HOCM patient, the signal separation algorithm and experimental verification of the CWDS-LVOT based on the MR-HOCM patient based on the FAST-ICA are completed, and the independence of the estimated source signal is measured using the duty rate. The LVOTG estimation is compared with the expert diagnosis results. Using sensitivity and specificity, the accuracy of the algorithm is evaluated, and an algorithm evaluation model based on the Subspace discriminant ensemble classifier is established. (3) the CWDS-LVOT feature parameters of the MR-HOCM patients are extracted and studied for the extraction of the CWDS-LVOT characteristic parameters of the MR-HOCM patients. Method: using the single degree of freedom model to extract the maximum frequency curve of CWDS-LVOT in MR-HOCM patients and extract the 8 characteristic parameters based on the maximum frequency curve extracted. The extracted feature parameters include the maximum systolic velocity S, the lowest end diastolic flow velocity D, the systolic and relaxation velocity ratio SD, the LVOTG value, the resistance index RI, the pulsation index PI, and the contraction. The spectrum width W and the systolic rise time T, and use the obtained characteristic parameters to evaluate the blood flow status. (4) the system interface design uses MATLAB GUI to design a simple analysis and identification system for MR-HOCM patients CWDS-LVOT. The operation interface is divided into LVOTG high estimation, feature extraction and diagnostic analysis report three most of the overestimation of.LVOTG In the decision section, the original MR-HOCM patient CWDS-LVOT was displayed, the source signal was estimated, the contrast map of the source signal and the CWDS-LVOT maximum frequency curve of the MR-HOCM patients after the correction of the flow rate and the corrected flow velocity and LVOTG value. The feature parameter extraction and the diagnosis analysis report part were used to compile the callback function of the corresponding control, and the corresponding clinical diagnosis was combined. The parameters and patient information are broken and the analysis report of the characteristic parameters and analysis results is generated in the form of WORD. In this paper, the MR-HOCM patient CWDS-LVOT is deeply studied and analyzed. A simple analysis and identification system for the flow velocity of CWDS-LVOT in the MR-HOCM patient is established to realize the flow velocity and the high LVOTG value for the CWDS-LVOT in the MR-HOCM patient. The estimation and correction of the MR-HOCM patients' CWDS-LVOT feature parameters are realized automatically, and the effectiveness and feasibility of the system are further verified by experiments.
【学位授予单位】：西华大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R542.2;TP391.41

【参考文献】