【作者】 范丽；

【导师】 肖湘生； 刘士远；

【作者基本信息】 第二军医大学 ， 影像医学与核医学， 2008， 博士

【摘要】 【研究背景】肺灌注的评价在临床上是非常重要的,检测局部灌注的异常改变有助于肺功能的评价。肺实质MRI由于肺组织内质子密度低,气体/组织界面大,磁敏感率不均匀,呼吸运动及心脏搏动伪影等因素,成像质量一直不是很理想。随着MRI成像技术的快速发展,肺实质MRI功能成像已成为现实。肺实质MRI功能成像包括肺通气成像和肺灌注成像两大类。MRI灌注成像是指利用快速扫描技术显示组织的微血管分布及血液灌注情况,提供组织的血流动力学信息,从影像学角度评估组织活力和功能的成像方法。与肺灌注的金标准核医学相比,MRI具有较高的空间分辨率,无辐射等优点,其价值正越来越受重视。MRI肺灌注成像的方法主要有:对比剂首过技术（first-pass contrast agenttechnique）,动脉自旋标记技术（arterial spin labeling,ASL）等。ASL是近几年新出现的MRI灌注成像方法,其最大的优点是不用静脉注射对比剂,而是以磁化标记的血管内自由流动的水质子作为内源性示踪剂来评价组织的灌注,降低了患者的检查费用和注射对比剂引起的潜在危险。ASL根据标记方式不同分为连续式（continuousarterial spin labeling,CASL）和脉冲式（pulsed arterial spin labeling,PASL） 2大类。本论文针对目前国际上肺功能MRI的研究热点,着重对血流敏感性的交替反转恢复（flowsensitive alternating inversion recovery,FAIR,属于PASL）序列参数进行了优化、评价了重力和呼吸对肺灌注的影响,在此基础上与3D动态增强MRI肺灌注进行对照研究,探讨ASL技术作为一种不用对比剂的非侵入性成像方法在肺实质MRI灌注成像中的临床应用价值。第一部分血流敏感性交替反转恢复序列在肺实质MRI灌注成像中的应用研究【研究目的】首先探讨血流敏感交替反转恢复序列（FAIR）作为一种不用对比剂的非侵入性灌注成像方法在肺实质MRI灌注成像中的可行性;为获得最佳的肺实质MRI灌注图像,优化其主要成像参数——反转恢复时间（inversion time,TI）,在此基础上评价重力和肺组织膨胀程度对肺灌注血流分布的影响。【材料与方法】应用GE 1.5T双梯度HD磁共振成像系统,先后对32例健康志愿者,用SSFSE-FAIR序列进行冠状面的肺实质MRI灌注成像。12例健康志愿者用于参数的优化研究:TI值分别取800ms、1000ms、1200ms、1400ms、1600ms。评价不同TI值时,反转脉冲标记前后肺实质信号强度变化率、主动脉信号强度变化率和相对肺灌注（rPBF）的信噪比（SNR）。10例健康志愿者评价重力对肺血流分布的影响:在仰卧位时,自背侧至腹侧（依次是P3、P6、P9、P12、P15）行5个冠状面的扫描,分别对5个冠状面的rPBF间进行方差分析,同一层面左/右肺rPBF间进行配对t检验,并对5个不同位置和rPBF间进行线性回归分析。另外10例健康志愿者用于肺组织膨胀程度对肺灌注影响的研究:分别在呼气末和吸气末屏气时进行冠状面扫描。分析在不同呼吸相时反转脉冲标记前/后双肺信号强度变化率（⊿SI%）、相对肺血流量（rPBF）的变化及扫描层面肺面积（Area）的变化情况。【结果】1.扫描参数的优化研究:①肺实质信号强度变化率:TI=800ms、1600ms时,P＞0.05;其余两两之间P＜0.05,且当TI=1000ms时,信号强度变化率最大。②主动脉信号强度变化率:TI=1400ms、1600ms时两两之间P＞0.05;其余两两之间P＜0.05,且当TI=1000ms时,信号强度变化率最大。③rPBF的SNR:TI=1000ms、1200ms、1400ms与1600ms时两两之间P＞0.05,其余两两之间P＜0.05。2.重力对肺灌注影响的研究:①在重力方向上存在灌注梯度,即5个冠状面的rPBF经方差分析均有统计学差异;经SNK-q检验得P12与P15间无统计学差异,其余两两之间均有统计学差异,rPBF由后至前是逐渐减小的。②在非重力方向上,肺灌注不存在灌注梯度,即同一冠状面,左、右肺rPBF之间无统计学差异。③线性回归分析得右肺的回归系数（RegressionCoefficients）是-4.98,左肺的回归系数是-5.16,即在仰卧位时自背侧向腹侧每增加1cm,右肺rPBF就减少4.98,左肺rPBF减少5.16。3.肺组织膨胀程度对肺灌注的影响研究:①不同呼吸相时⊿SI%之间有明显的统计学差异（右肺P=0.0215,左肺P=0.0084）,呼气末的⊿SI%明显高于吸气末。②不同呼吸相时双肺rPBF间均有统计学差异（右肺P=8.92×10-5,左肺P=0.0002）,呼气末的rPBF明显高于吸气末。③不同呼吸相扫描层面的Area间也有明显的统计学差异（右肺P=2.94×10-5,左肺P=0.0005）,吸气末面积明显大于呼气末面积。【结论】FAIR用于肺实质灌注成像是可行的。当TI=1000ms时,肺实质和主动脉的信号强度变化率最大,信噪比较好,可以获得较好的灌注图像。FAIR对重力方向的灌注梯度及不同呼吸相时肺灌注间的差异是比较敏感的,所以用FAIR技术进行肺灌注检查时,患者的体位和屏气时相是非常重要的。熟悉重力所致的肺血流分布的不均匀性和不同呼吸相时肺灌注的差异,在检查时就可以人为的改变受检者体位,将感兴趣区置于重力依赖性区域,并在呼气末屏气可以提高灌注缺损的检出率。第二部分三维并行采集动态增强MRI在肺实质局部灌注中的应用研究【研究目的】评价3D并行采集MRI技术对健康志愿者和肺部疾病患者的肺实质局部灌注成像的可行性和可重复性。【材料与方法】应用GE 1.5T磁共振成像系统,对10例健康志愿者,47例肺部疾病患者（32例原发性肺癌,5例炎性病变,10例肺动脉栓塞）行MRI成像。用高时间分辨率的3D并行采集梯度脉冲序列（GE商品名:LAVA,liver acquisition with volume acceleration即肝脏快速容积采集序列）在注射对比剂前先进行冠状面蒙片扫描;之后用高压注射器经肘静脉以3ml/s的注射速率,注射30ml对比剂（Gd-DTPA）4s后,启动3D冠状容积采集。在注射对比剂结束后,以相同的注射速率追加注射相同量的生理盐水。将注射对比剂后每个期相的图像与蒙片进行减影得到灌注图像。评价所有受检者肺灌注图像的均匀度,是否存在灌注异常区域,若存在灌注异常区域,则计算灌注异常区与正常区的信号强度之比R_Sl。应用单样本t检验进行统计学分析,检验水准α=0.05。【结果】10例正常志愿者均得到了高分辨率、运动伪影很小的灌注图像,未见灌注缺损区。在24s屏气时间里,DCE-MRI可以清楚的显示肺实质灌注情况和肺血管结构。22例肺部疾病患者（10例肺动脉栓塞、12例侵犯肺血管的肺癌）共出现了24个楔形或三角形的灌注缺损区。10例PE共有12个灌注缺损区,其中1例是双侧多发PE有3个灌注缺损区,其余PE均发现了1个灌注缺损区。12例肺癌均侵犯了邻近的肺动脉,均在其远端相应的供血区出现了灌注缺损。R_SI经单样本t检验得,t=-24.74,P＜0.05,具有明显的统计学意义,即灌注缺损区与正常灌注区的信号强度间有明显的统计学差异。其余25例肺部疾病患者（20例肺癌未侵犯血管、5例炎性病变）在对比剂首过肺循环肺实质强化达峰值时,病灶局部均呈低信号改变,即所有病灶与高灌注的肺组织相比均呈低灌注。【结论】3D并行采集MRI技术可在单次屏气状态下完成动态多期扫描,减少了呼吸运动对后期图像处理的影响,同时获得了全肺的容积灌注数据。进行MR肺灌注成像的同时,也完成了MR肺血管造影,两者结合更有利于观察肺栓塞等肺部血流异常疾患。对MR肺灌注图像采用半量化分析方法,不但可以提高对肺灌注异常区域的检出率,而且能达到对异常区域病变性质分析诊断的效能。第三部分SSFSE-FAIR与3D DCE-MRI在肺实质MRI灌注成像中的对照研究【研究目的】评价SSFSE-FAIR在正常志愿者、肺癌、肺动脉栓塞患者肺实质灌注成像中的应用,并将其与高时间分辨率的3D DCE-MRI进行对照研究。【材料与方法】应用GE 1.5T磁共振成像系统,对10例健康志愿者和30例肺部疾病患者（10例肺动脉栓塞、20例肺癌）先后进行SSFSE-FAIR与3D DCE-MRI扫描。评价肺实质灌注图像的均匀度,若出现灌注异常,则计算灌注异常区域与正常灌注区域的信号差异（⊿SI=SI normal-SI abnormal/SI normal）。应用配对t检验进行统计学分析,检验水准α=0.05。【结果】FAIR（TI=1000ms）与DCE-MRI（LAVA）均可获得图像质量较好的灌注图像。正常志愿者肺实质灌注是均匀的,无灌注缺损区。10例肺动脉栓塞和12例侵犯肺动脉的中央型肺癌患者均可见到楔形或三角形的灌注缺损区。⊿SI在FAIR与DCE-MRI两方法中无明显的统计学差异（P＞0.05）。其余8例周围型肺癌患者,病变区域在FAIR与DCE-MRI两种方法中相对于正常肺实质均为低灌注。【结论】FAIR与DCE-MRI应用于肺实质灌注成像是可行的且二者具有一致性。对于肺癌或肺动脉栓塞引起的灌注异常有较高的诊断精确性。FAIR作为一种非侵入性的方法可以获得与DCE-MRI相同的功能与诊断信息。更多还原

【Abstract】 【Background】Assessment of pulmonary perfusion is very important in clinical practice.Changes in regional pulmonary perfusion can be observed in a variety of lung diseases,such as pulmonary embolism,chronic obstructive pulmonary disease.Pulmonary parenchyma magnetic resonance imaging has been hampered by several factors,such as intrinsic low spin density,heterogeneous magnetic susceptibility,respiration and cardiac motion, pulmonary blood flow and molecular diffusion.All these factors attribute to the low quality of pulmonary parenchyma imaging.But with the fast development of MRI techniques, pulmonary parenchyma MRI has come into truth.Compared to nuclear medicine scintigraphy,MRI has several advantages,such as nonionization exposure and high spatial resolution that influence the quality of images.Pulmonary parenchyma MRI includes pulmonary ventilation imaging and pulmonary perfusion imaging.Methodologically,Perfusion imaging includes two basic approaches: arterial spin labeling（ASL） and dynamic contrast-enhanced（DCE） imaging.Arterial spin labeling is a relatively new technique for MRI perfusion imaging.It does not require injection of contrast agents,and uses magnetically labeled blood water as an endogenous, freely diffusible tracer;hence it is completely noninvasive.ASL can be subdivided into steady state and pulsed.In this study,a pulsed ASL technique:flow sensitive alternating inversion recovery（FAIR） was used.In this study,we emphasized on the evaluation of feasibility of pulmonary parenchyma perfusion imaging with FAIR,the optimization of TI value in order to acquire high quality perfusion images and the assessment the influence of gravity and lung inflation on pulmonary perfusion.Based above research,we evaluated the potential clinical application value of FAIR prospectively by comparing FAIR with 3D DCE-MRI.Part OnePulmonary Parenchyma Magnetic Resonance Perfusion Imaging with FAIR【Objective】To evaluate the feasibility of pulmonary parenchyma perfusion imaging with FAIR,optimize TI value of FAIR in order to acquire high quality perfusion images and evaluate two influence factors on pulmonary perfusion,such as gravity and lung inflation.【Materials and methods】A total of 32 healthy volunteers were undergone SSFSE-FAIR imaging.12 volunteered participated in optimization TI study.Five different TI values were 800ms,1000ms,1200ms,1400ms and 1600ms,respectively.The changes of signal intensity between control and tag images in lung and aorta were evaluated,signal to noise ratio（SNR） of perfusion images was also evaluated with different TI values.Ten volunteers were included in the gravity-dependent perfusion study.Five coronal slices at an interval of 3cm from dorsal to ventral（labeled as P3,P6,P9,P12,P15,sequently） were obtained when the volunteers performed breath holding on end expiration and the relative pulmonary blood flow（rPBF） was measured.Coronal perfusion-weighted images were obtained at different respiratory phases from the rest of 10 healthy volunteers using FAIR sequence.The change of tagging efficiency of pulmonary parenchyma（⊿SI%）, pulmonary blood flow and area of the scanning slice of different respiratory phases were analyzed.【Results】1.Optimization TI study:①signal intensity change of lung:no statistic difference between 800 ms and 1600 ms（P＞0.05）,there were significant statistic differences of any other two TI values（P＜0.05）.the change was maximum at 1000ms;②signal intensity change of aorta:no statistic difference between 1400 ms and 1600 ms （P＞0.05）,there were significant statistic differences of any other two TI values （P＜0.05）.the change is maximum at 1000ms;③SNR of perfusion images:no statistic differences among 1000ms,1200ms,1400ms and 1600ms（P＞0.05）,there were significant statistic differences of any other two TI values（P＜0.05）.2.Gravity-dependent study: Along the direction of gravity,a gradient was visually perceived as a vertical increase in rPBF.There were significant statistic differences in rPBF between any two coronal planes except that between P 12 and P 15.In supine position,regression coefficients of right and left lung were—4.98 and—5.16,respectively.This means that rPBF decreased 4.98 （right） and 5.16（left） for each centimeter above the dorsal.No statistical difference was seen between ROIs placed along iso-gravitational plane.3.Lung inflation study:①Significant differences was found in⊿SI%between different respiratory phases(right lung P=0.0215,left lung P=0.0084,with that at end expiration greater than that at end inspiration.②Significant differences was also found in pulmonary blood flow at different respiratory phases(right lung P=8.92×10^-5,left lung P=0.0002),with that at end expiration higher than that at end inspiration.③The area of the scanning slice was also significantly different at different respiratory phases(right lung P=2.94×10^-5,left lung P=0.0005),with that at end inspiration larger than that at end expiration.【Conclusion】Pulmonary perfusion with FAIR is feasible.1000ms TI was the best choice,which generated the highest lung/aorta tagging efficiency and second highest PBF SNR.Both gravity and respiratory phase are important determinants of pulmonary perfusion heterogeneity.FAIR is sensitive to demonstrate gravity- and respiratory phase- dependent differences in lung perfusion.Positioning the patient so that the area of interest is down-gravity and asking patient to hold breath on end expiration may help in detection of perfusion defects.Part TwoPulmonary Parenchyma Perfusion Imaging with 3D-DCE MRI【Objective】To assess feasibility of pulmonary parenchyma perfusion with 3D dynamic contrast-enhanced（DCE） imaging in healthy volunteers and in patients.【Materials and methods】Ten healthy volunteers and 47 patients（32 with lung cancer,5 with inflammation,and 10 with pulmonary emboli） were included in this study.A time-resolution three dimensional gradient-echo pulse sequence with parallel acquisition was used for DCE-MRI.Firstly,mask images with LAVA in a coronal orientation were scanned before injection of contrast agent.Then,six phases of 3D coronal volume were acquired consecutively,starting 4 seconds after the injection.All injections were performed with an automatic power injector.30ml Gd-DTPA was injected followed by 30ml saline flush into the antecubital vein at a rate of 3ml/s.Mask images were subtracted from each phase for a pure perfusion image.The homogeneity of perfusion of volunteers was assessed.In case of perfusion abnormality,the contrast between normal lung and perfusion defects was quantified by calculating the signal intensity difference ratio(R_SI between the perfusion abnormalities and the opposite side normal lung.【Results】The signal intensity of perfusion images of healthy volunteers was homogeneous.Pulmonary parenchyma perfusion as well as pulmonary vasculature was depicted during a 24-s breath-hold with DCE-MRI.24 wedged-shaped or triangle perfusion defect or reduced regions were visualized in 22 patients including 10 pulmonary embolisms and 12 lung cancers infiltrating the pulmonary artery.There was significant difference in R_SI（t= -24.74,P 0.05）.In the other 25 cases,all lesions showed hypo-intensity when pulmonary parenchyma was enhanced to peak during the first-pass of contrast agent through the pulmonary circulation.【Conclusion】3D parallel acquisition could obtain multiphases of dynamic volume scanning during a single breath-hold and minimize the respiratory artifaction.It could acquire not only perfusion imaging,but also the pulmonary angiography.The combination of perfusion imaging and angiography could be easier to assess the pulmonary blood abnormal diseases,such as pulmonary embolism. Semiquantitative analysis could improve the detection rate of perfusion abnormal. Pulmonary perfusion with DCE-MRI is feasible.Part ThreeAssessment of Pulmonary Parenchyma Perfusion with FAIR in Comparison with DCE-MRI【Objective】The aim of this study was to assess pulmonary parenchyma perfusion with flow-sensitive alternating inversion recovery（FAIR） in comparison with 3D dynamic contrast-enhanced（DCE） imaging in healthy volunteers and in patients with pulmonary embolism or lung cancer.【Materials and methods】FAIR imaging with the optimized TI was performed followed by DCE-MRI on the 10 healthy volunteers and 30 patients with pulmonary embolism（10 cases） or lung cancer（20 cases）.The homogeneity of FAIR and DCE-MRI perfusion was assessed.In the cases of perfusion abnormality,the contrast between normal lung and perfusion defects was quantified by calculating a normalized signal intensity ratio.【Results】In the healthy volunteers,the signal intensity of perfusion images acquired with both FAIR and DCE-MRI was homogeneous.Wedged-shaped or triangle perfusion defects were visualized in 10 pulmonary embolisms and 12 lung cancer cases.There was no significant statistical difference in signal intensity ratio between FAIR and DCE-MRI（P＞0.05）.In the rest of 8 lung cancers,all the lesions showed low perfusion against the higher perfused pulmonary parenchyma in both FAIR and DCE-MRI.【Conclusion】Pulmonary parenchyma perfusion imaging with FAIR and DCE-MRI was feasible and consistent,which has high diagnosis accuracy for the detection of perfusion abnormalities caused by pulmonary embolism or lung cancer.FAIR as a noninvasive technique could obtain the similar functional information,comparing with DCE-MRI.更多还原