狼疮性肾炎患者的功能性磁共振成像研究

发布时间：2018-03-18 03:19

本文选题：血氧水平依赖　切入点：扩散加权成像　出处：《天津医科大学》2017年博士论文　论文类型：学位论文

【摘要】：第一部分狼疮性肾炎肾脏血氧水平依赖磁共振成像研究【目的】狼疮性肾炎患者恰当治疗和合理预后评估依赖于对其肾脏病生理状态实时精确的评估。血氧水平依赖磁共振成像技术可提供组织内氧合状态的信息。我们旨在利用此成像技术获取的肾脏氧合状态图像构建利用磁共振指标预测肾脏病理类型的算法模型和探索病理特征与氧合状态的关联。【方法】选取经2012年国际狼疮协作组诊断标准确诊的狼疮肾炎患者和年龄性别匹配成人作为研究对象。采用GE Discovery 750 3.0T磁共振成像系统对临床确诊的狼疮性肾炎患者和健康对照人群采集血氧水平依赖磁共振图像,并利用功能重建Functool和R2*map后处理软件生成肾脏R2*值图像。利用数值分析软件Math Works MATLAB2015a测量肾脏感兴趣区R2*值水平和图像纹理特征中灰度共生矩阵特征值。通过统计和数据挖掘分析构建预测肾脏病理类型的决策树、Fisher线性判别方程和Logistic回归方程三种算法模型,并对比其三者的诊断效能。建立和绘制狼疮肾性肾炎患者肾脏R2*值与不同病理特征之间的多重对应分析关联图。【结果】利用R2*值数据特征参数构建的决策树算法模型对预测肾脏病理类型的受试者工作曲线下面积高于线性判别方程模型(0.765 vs 0.629,P0.001)和Logistic回归模型(0.765 vs 0.662,P0.001)。R2*值图像的灰度共生矩阵特征显示,相关和能量两指标在不同肾脏病理类型间存在差异。利用这两项指标构建的Fisher线性判别方程对病理类型预测的总体正确率可达到77.8%。肾脏R2*值水平显示,V型狼疮肾脏R2*值(24.291±2.482)高于相应的III型狼疮(19.576±2.854,p0.001)和IV型狼疮(20.284±3.857,p0.001)。多重对应分析显示较高的R2*值与较轻的肾脏小管和间质损伤相和较高的年龄相关联。【结论】血氧水平依赖磁共振成像技术可用于评价肾脏组织氧合状态。利用血氧水平依赖磁共振指标和图像特征有可能建立预测肾脏病理类型的算法模型。狼疮性肾炎肾脏组织内氧合状态与其病理损伤的程度之间可能存在较复杂的非线性关联。第二部分狼疮性肾炎肾脏扩散加权磁共振成像研究【目的】狼疮性肾炎肾脏病理改变可引起细胞和组织间水分子转运的异常。扩散加权磁共振成像可提供细胞和组织内水分子的运动特征信息。我们旨在利用此技术获取的磁共振图像,探索健康人群和狼疮性肾炎患者肾脏组织内水分子的运动特征,并借此评价狼疮性肾炎肾脏的病生理状态。【方法】选取经2012年国际狼疮协作组诊断标准确诊的狼疮肾炎患者和年龄性别匹配成人作为研究对象。采用GE Discovery 750 3.0T磁共振成像系统对临床确诊的狼疮性肾炎患者和健康对照人群采集扩散加权磁共振图像,并利用功能重建Functool和MADC后处理软件生成肾脏ADC值、D值、D*值和f值图像。利用MATLAB数值分析软件分别测量肾脏感兴趣区ADC值、D值、D*值和f值水平。比较健康对照人群与狼疮肾炎患者以及不同病理类型狼疮肾炎患者之间肾脏各指标间的差异。分析扩散加权成像指标与不同病理特征之间的多重对应关联。【结果】健康对照人群肾脏ADC值水平皮质区高于髓质区(2.393±0.085 vs2.358±0.067′10-3mm2/s,P=0.002),肾脏f值水平皮质区高于髓质区(23.949±0.321 vs 23.530±0.735%,P0.001),而D值和D*值皮髓质区无差异。狼疮组中D值皮质区水平高于髓质区水平(1.677±0.314 vs 1.597±0.348′10-3mm2/s,P=0.03),而D*值皮质区水平低于髓质区水平(22.391±1.979 vs 23.079±1.333′10-3mm2/s,P0.001)。而ADC值和f值两项指标皮质区与髓质区水平无统计学差异。狼疮肾炎组皮质区和髓质区内,除D*值指标高于健康对照组外,其ADC值、D值和f值均低于健康对照组。不同病理类型的狼疮性肾炎之间其扩散加权指标也存在不同程度的差异。多重对应分析显示,扩散加权成像指标水平与不同病理损害严重程度间存在关联。【结论】扩散加权磁共振成像技术可用于评价肾脏组织内水分子的运动特征,狼疮性肾炎肾脏组织内水分子的运动特征发生改变。通过测量肾脏组织内水分子运动指标有可能了解到肾脏组织病理损伤的类型和程度。
[Abstract]:The first part of lupus nephritis renal blood oxygen level dependent magnetic resonance imaging study [Objective] in patients with lupus nephritis and proper treatment and reasonable prognosis depends on the physiological state of kidney disease can provide real-time and accurate evaluation. Tissue oxygenation status information of blood oxygen level dependent magnetic resonance imaging. We aimed to use the oxygen to get kidney the use of magnetic resonance imaging technology to construct index to predict the type of renal pathology algorithm model and to explore the pathological features and oxygenation state image. [method] the 2012 international cooperation group of lupus diagnostic criteria for diagnosis of lupus nephritis patients and age and gender matched adults as the research object. Using GE Discovery 750 3.0T magnetic resonance imaging system clinical diagnosis of lupus nephritis patients and healthy control group collected blood oxygenation level dependent magnetic resonance images, and work hard Can Functool and R2*map reconstruction postprocessing software to generate R2* in the kidney of binary images. The numerical analysis software Math Works MATLAB2015a to measure the kidney region of interest R2* level and image texture feature of gray level co-occurrence matrix eigenvalue. Through statistical analysis of data mining and decision tree construction in predicting renal pathological types, Logistic equation and the regression equation of three algorithms Fisher linear discriminant model, and compare the diagnostic efficacy of three. The establishment and analysis of association graph and multiple correspondence between different pathological features of renal R2* rendering in patients with lupus nephritis renal values. [results] using R2 * values to construct characteristic parameter algorithm of decision tree model to predict the type of renal pathology of the ROC was under the above linear discriminant equation model (0.765 vs 0.629, P0.001) and Logistic regression model (0.765 vs 0.662, P0.001).R2* image gray scale co-occurrence matrix Array characteristics show that exist in different types of renal pathology correlation and energy index difference. Two criteria for pathological types overall prediction accuracy can reach 77.8%. level of renal R2* display using Fisher linear construction of the two indicators, V lupus renal R2* values (24.291 + 2.482) higher than the corresponding (III lupus 19.576 + 2.854, p0.001) and IV lupus (20.284 + 3.857, p0.001). Multiple correspondence analysis showed that the higher R2* value and lower renal tubule and interstitial injury and associated with higher age. [Conclusion] the blood oxygenation level dependent magnetic resonance imaging can be used to evaluate the renal tissue oxygenation using oxygen. Level dependent magnetic resonance parameters and image features may predict the algorithm model of renal pathological types. There is a complex between the degree of oxygenation and pathological injury in renal tissue of lupus nephritis in May Nonlinear correlation. The second part lupus nephritis renal diffusion weighted magnetic resonance imaging study [Objective] lupus nephritis renal pathological changes can cause cell and interstitial water molecules transport anomaly. The information diffusion weighted magnetic resonance imaging can provide cellular and tissue water molecular motion features. We aimed to magnetic resonance images obtained by this technique to explore, movement characteristics of healthy people and patients with lupus nephritis kidney water molecules, and to evaluate the physiological state of disease of lupus nephritis kidney. [method] the 2012 international cooperation group of criteria for the diagnosis of lupus diagnosed lupus nephritis patients and age and gender matched adults as the research object. Using GE Discovery 750 3.0T magnetic resonance imaging the system for clinical diagnosis of lupus nephritis patients and healthy control group acquisition of diffusion weighted magnetic resonance images, and the use of function The Functool and MADC postprocessing software to generate the kidney ADC value, D value, D* value and F value were measured in kidney images. The value of ADC in the region, using MATLAB numerical analysis software D value, D* value and F value level. Compared to healthy control group and lupus nephritis patients and between patients with different pathological types of renal inflammation in each the difference between the index of kidney. Multiple correspondence analysis correlation between diffusion weighted imaging and pathological characteristics of different indexes. [results] in kidney ADC level medulla cortex than healthy control (2.393 + 0.085 + 0.067 'vs2.358 10-3mm2/s, P=0.002), the F level of kidney cortex than medulla (23.949 + 0.321 vs 23.530 + 0.735%, P0.001), while D and D* values had no difference in the corticomedullary zone. Cortex medulla level is higher than the level of D in lupus group (1.677 + 0.314 vs 1.597 + 0.348 10-3mm2/s', P=0.03), while the D* value level is lower than the medulla cortex The level of (22.391 + 1.979 vs 23.079 + 1.333 10-3mm2/s', P0.001). ADC and F values of two indexes of cortex and medulla level was not statistically significant. Lupus nephritis group, cortex and medulla area, in addition to the value of D* index is higher than the healthy control group, the ADC value, D value and F value lower than the healthy control group. Between different pathological types of lupus nephritis and its diffusion weighted index are different. Multiple correspondence analysis showed that the diffusion weighted imaging indexes with different pathological damage associated severity. [Conclusion] diffusion weighted magnetic resonance imaging technique can be used for the evaluation of kidney tissue motion characteristics of water molecules. The movement characteristics of lupus nephritis kidney water molecular change. To understand the type and degree of kidney injury by measuring the kidney index movement of water molecules is possible.

【学位授予单位】：天津医科大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：R445.2;R593.242

【相似文献】