【作者】 张月；

【导师】 程敬亮；

【作者基本信息】 郑州大学 ， 放射影像学（专业学位）， 2024， 硕士

【摘要】 背景与目的缺血性脑卒中作为全球范围内的重大公共卫生问题,在发达国家和发展中国家中均导致高发病率和死亡率,同时也是导致残疾的第三大原因,给患者的生活质量带来持续的负面影响。因此,早期诊断和治疗缺血性脑卒中对改善患者预后至关重要。磁共振成像(magnetic resonance imaging,MRI)技术在缺血性脑卒中的诊断和康复研究中扮演着关键角色。其中,定量磁共振成像(quantitative magnetic resonance imaging,qMRI)技术能够提供敏感的组织生理变化指标,为缺血性脑卒中的早期诊断和治疗提供重要的组织信息。然而,缺血性脑卒中所引起的组织坏死和水肿可能导致颅内高压,从而使患者出现不自主运动,这给qMRI技术的广泛应用带来了挑战,尤其是在需要长时间扫描且易受运动伪影影响的qMRI中。因此,迫切需要探索更快、更准确和抗运动的成像技术。在缺血性脑卒中后导致成年人长期残疾的因素中,皮质脊髓束(corticospinal tract,CST)完整性的损伤起着重要作用。在缺血性脑卒中后,神经细胞体或近端轴突受损后会导致远端白质的神经纤维束发生变性,即华勒氏变性(Wallerian degeneration,WD),并且CST是缺血性脑卒中后最容易发生WD的神经纤维束。因此,早期检测缺血性脑卒中后CST的变化是反映运动功能状况的重要因素。然而,大部分患者在常规MRI上很难检测和量化缺血性脑卒中后CST发生的变性,即看起来正常的皮质脊髓束(normal-appearing corticospinal tract,NA-CST)。扩散张量成像技术(diffusion tensor imaging,DTI)可以无创显示脑白质纤维走行,定量显示微结构异常改变,其中分数各向异性(fractional anisotropy,FA)指标可以衡量白质纤维束完整性,为脑卒中后的早期诊断提供有价值的信息。然而,DTI技术以及更高阶的扩散模型因对硬件要求高,扫描时间长以及图像重建复杂等限制而难以在临床诊断中得以广泛应用。此外,基于纵向弛豫时间(T1)和横向弛豫时间(T2)的qMRI技术也能够提供敏感的组织生理变化指标,为缺血性脑卒中的早期诊断和治疗提供详细的影像信息。它能够根据组织内的水分子及脂质含量、分子成分、神经元密度等组织特性显示出不同的定量值,并且在检测组织完整性方面具有高度敏感性,成为揭示与疾病相关的细微生物学变化的新途径。本研究采用的基于多重叠回波分离(multiple overlapping-echo detachment,MOLED)成像的超快速单扫描定量T2 mapping技术,能够在150 ms左右获得一层大脑切片的定量T2图,与其他定量技术一样,该技术理论上不受设备类型、扫描参数及操作方式的影响,能够提供客观且可重复的微小分子变化信息,同时具有超快速、抗运动的特性,在缺血性脑卒中患者的图像采集中展现出显著优势。本研究的目的是使用基于MOLED的T2 mapping技术在探索缺血性脑卒中后检测NA-CST的损伤及运动功能预后中的应用价值。材料与方法1、验证试验:首先在10个装有不同浓度的氯化锰溶剂的试管模型和五名健康志愿者中验证MOLED技术的准确性、可重复性和抗运动的特性,在模型试验中,使用多扫描自旋回波序列(multi-sectionmulti-echo spin-echo sequence,MESE)作为参考测量的金标准序列;在健康志愿者实验中,进行五次无运动重复扫描以及一次运动状态下扫描,并分别提取七对不同脑区的感兴趣区域(regions of interest,ROIs)的T2值。在运动试验中,使用传统的T2加权成像技术(T2-weighted imaging,T2WI)扫描得到的图像作为对照。2、研究对象:本研究共纳入79名亚急性期前循环缺血性脑卒中患者,使用西门子Prisma3.0TMR扫描仪对所有患者进行扫描,包括常规MRI扫描序列和基于MOLED T2序列。使用美国国立卫生研究院卒中量表中提取的第5项和第6 项的上下肢运动功能评分(National Institute of Health Stroke Scale motor index,mNIHSS)对所有患者进行运动功能障碍的严重程度的评估,随后将患者按mNIHSS 评分高低分为 mNIHSS Ⅰ(mNIHSS,0 分):22 例、mNIHSS Ⅱ(mNIHSS,1～4分):45例;mNIHSS Ⅲ(mNIHSS,5～8分):12例的运动功能障碍严重程度不同的三组患者。3、数据处理:使用 MRIcron(www.sph.sc.edu/comd/roden/mricron)在 T2 FLAIR图像上逐层手动绘制由缺血灶和白质高信号(white matter hyperintensities,WMH)组成的病变掩膜,并测量缺血灶体积(ischemic lesion volume,ILV);在FSL(FMRIB 软件库,fsl.ffmrib.ox.ac.uk/fsl/fslwiki/Atlases)中对数据进行配准处理,使用约翰霍普金斯大学白质束成像图集(Johns Hopkins University white matter tractography atlas,JHU-CST)作为CST束的标准模板,剔除脑脊液(cerebrospinal fluid,CSF)及病变掩膜和概率值小于20%的CST以得到T2 mapping空间的NA-CST;分别提取缺血灶同侧及对侧NA-CST的平均T2值并计算双侧的T2比值(T2 ratio,rT2)。4、统计分析:使用SPSS软件(23.0版,IBM Corp,Armonk,NY)进行统计学分析。在技术验证试验中,采用Pearson相关性和Bland-Altman分析法,并通过正态性检验、变异系数(coefficient of variation,CoV)、组内相关系数(intra-group correlation coefficient,ICC)及其 95%置信区间(confidence interval,CI)来测试技术的准确性及可重复性。随后通过Bland-Altman分析来检验在有无运动状态下扫描结果的一致性情况。在缺血性脑卒中后NA-CST损伤的研究中,首先确认参数的正态分布情况,随后使用Wilcoxon符号秩检验比较所有患者及运动功能障碍严重程度不同的三组患者缺血灶同侧与对侧NA-CST的T2值,采用方差分析(analysis of variance,ANOVA)和Bonferroni校正对运动功能障碍严重程度不同的三组患者的ILV进行两两比较。使用Spearman相关分析探讨ILV和rT2值与运动功能障碍的严重程度之间的相关性,随后分析ILV与缺血灶同侧NA-CST的T2值之间的相关性,并使用rT2和ILV作为自变量的有序Logistic回归模型来预测缺血性脑卒中后运动功能障碍的严重程度。最后选择了对预测运动功能障碍严重程度有意义的rT2,计算最佳临界值和受试者工作特征曲线(receiver operating characteristic,ROC)的曲线下面积(area under the curve,AUC)。结果1、验证试验结果显示:分别基于MOLED和MESE获得的定量T2值之间具有很强的正相关性,皮尔逊相关系数(Pearson correlation coefficient,PCC)为0.9995(P<0.0001)。此外,两种方法具有高度的一致性,平均差(meandifference,Meandiff)=4.20%,差值标准差(standard deviation of difference,SDdiff)=1.71%。健康志愿者验证试验结果表明基于MOLED获得的T2值在每个志愿者的所有重复扫描中都具有极佳的可重复性。在前五次无运动的重复扫描中,所有ROIs均表现出良好的一致性,CoV在0.47到2.74之间。此外,运动实验显示,尽管常规序列受头部运动影响产生伪影,但基于MOLED获得的T2 mapping能够避免运动影响,获得高质量图像。Bland-Altman分析结果表明,无论是否存在运动,T2值的测量均无系统性偏差,Meandiff为0.28%,SDdiff为1.34%。2、缺血性脑卒中患者的临床资料:本研究共收集79名前循环缺血性卒中患者(26名女性,53名男性)平均年龄是(56.67± 11.88)岁,起病后到做MRI时间为18(8～38)天。本研究中,按照运动障碍的严重程度将患者分为三组:mNIHSSⅠ组22例;mNIHSS Ⅱ组45例;mNIHSS Ⅲ组12例。三组患者之间在性别、年龄分布及心血管危险因素方面的差异均无统计学意义。3、缺血性脑卒中后缺血灶同侧及对侧NA-CST的T2值比较:在所有患者中,缺血灶同侧NA-CST的平均T2值明显高于对侧[T2同侧:70.05(67.68～73.84);T2对侧:68.76(66.13～71.29);P<0.001]。在mNIHSSⅠ组患者中,缺血灶同侧和对侧NA-CST的平均T2值无明显差异[T2同侧:68.86(67.27～72.60);T2对侧:69.50(67.36～72.96);P=0.485],在mNIHSS Ⅱ组患者中,缺血灶同侧NA-CST的平均 T2值明显高于对侧[T2同侧:70.60(67.36～73.84);T2对侧:68.45(65.55～71.17);P<0.001],对于mNIHSS Ⅲ组患者,缺血灶同侧NA-CST的平均T2值同样明显高于对侧[T2同侧:70.75(68.49～80.07);T2对侧:69.11(66.58～70.57);P=0.002]。4、三组不同运动功能障碍严重程度的患者ILV的比较:mNIHSS Ⅰ组和mNIHSS Ⅱ组患者之间的 ILV 没有显著差异[ILVⅠ:1910(310～3400);ILVⅡ:1640(570～6270);P>0.05]。然而,在 mNIHSS I 组和 mNIHSS Ⅲ组患者之间[ILVⅠ:1910(310～3400);ILVⅢ:5740(2250～17780);P=0.014]以及 mNIHSSⅡ 和mNIHSS Ⅲ之间[ILVⅡ:1640(570～6270);ILVⅢ:5740(2250～17780);P=0.045]观察到ILV存在统计学差异。5、rT2及ILV与运动功能障碍严重程度的相关性分析:研究结果显示,rT2与运动功能障碍的严重程度呈正相关(r=0.575,P<0.001),rT2值越高,相应运动功能障碍的程度越严重。然而,尽管ILV与运动功能障碍的严重程度相关,但相关性较弱(r=0.290,P=0.010)。此外,研究还发现ILV与缺血灶同侧NA-CST的平均T2值不存在相关性(r=0.034,P=0.763)。6、rT2对缺血性脑卒中后运动功能的诊断效能:以rT2和ILV为自变量,运动功能障碍的严重程度为因变量的有序Logistic回归模型分析结果显示rT2对预测运动功能障碍严重程度有意义,当rT2的临界点在1.013时,有助于判断缺血性脑卒中后是否会发生运动功能障碍,灵敏度为80.70%,特异度为81.82%,AUC为0.883,误差为0.04。结论1、基于MOLED的T2 mapping技术具有超快速、抗运动以及准确且可重复的特性,在缺血性脑卒中患者的图像采集中具有一定的优势。2、基于MOLED的T2mapping技术能够检测到缺血性脑卒中后NA-CST的损伤,并初步评估缺血性脑卒中引起的运动功能障碍的预后情况。更多还原

【Abstract】 Background and PurposeIschemic stroke,as a significant public health issue globally,results in high incidence and mortality rates in both developed and developing countries.It is also the third leading cause of disability,leading to sustained negative impacts on patients’quality of life.Therefore,early diagnosis and treatment of ischemic stroke is essential to improve patient prognosis.Magnetic resonance imaging(MRI)techniques play a key role in ischemic stroke diagnosis and rehabilitation studies.Among them,quantitative magnetic resonance imaging(qMRI)provides sensitive indicators of tissue physiological changes,which provides important tissue information for early diagnosis and treatment of ischemic stroke.However,tissue necrosis and edema induced by stroke may lead to intracranial hypertension,which may cause patients to experience involuntary movements,posing a challenge to the widespread application of qMRI techniques,especially in qMRI sequences that require long scanning times and are susceptible to motion artifacts.Therefore,there is an urgent need to explore faster,more accurate,and motion-resistant imaging techniques.In the factors leading to long-term disability in adults after ischemic stroke,damage to the integrity of the corticospinal tract(CST)plays a crucial role in the development of post-stroke motor impairments After stroke,the degeneration of nerve fiber bundles in the distal white matter due to damage to neuronal cell bodies or proximal axons is referred to as Wallerian degeneration(WD),and the CST is the nerve fiber tract most susceptible to WD after stroke.Therefore,early detection of microstructural changes in the CST after stroke is an important factor in reflecting the status of motor function.However,it is challenging to detect and quantify the degeneration of the CST after stroke using conventional MRI,which is referred to as the normal-appearing corticospinal tract(NA-CST).Diffusion tensor imaging(DTI)can noninvasively show the white matter fiber alignment and quantify the microstructural abnormalities,of which the fractional anisotropy(FA)metric can measure the integrity of the white matter fiber tracts,providing valuable information for early diagnosis after stroke.However,DTI techniques and higher-order diffusion models are challenging to be widely applied in clinical practice due to their high hardware requirements,long scanning times,and complex image reconstruction processes.Furthermore,qMRI techniques based on longitudinal relaxation time(T1)and transverse relaxation time(T2)can also provide sensitive indicators of tissue physiological changes,offering detailed imaging information for the early diagnosis and treatment of ischemic stroke.The technique can show different quantitative values according to the water molecule and lipid content,molecular composition,neuron density,and other tissue properties,and is highly sensitive in detecting the integrity of tissues,and is now has begun to be applied in the study of white matter fiber bundles as a new way to reveal disease-related fine microbiological changes.Recently,the ultrafast single-scan T2 mapping technique based on multiple overlapping-echo detachment(MOLED)imaging is capable of obtaining quantitative T2 values for one layer of brain slices in about 150 ms,which,like other quantitative techniques,is theoretically independent of the type of equipment,scanning parameters and operation mode,can provide objective and reproducible information about small molecular changes,and has ultrafast and motion-resistant properties,demonstrating significant advantages in image acquisition of stroke patients.This study aimed to explore the value of detecting damage to the NA-CST after ischemic stroke and its application in the prognosis of motor function using MOLED-based quantitative T2 mapping.Materials and Methods1.Validation Experiment:The accuracy,repeatability,and motion robustness of the MOLED technique were first validated in ten test tube models containing MnCl2 solution with different concentrations and five healthy volunteers.In the model experiments,a multi-section multi-echo spin-echo sequence(MESE)was used as the gold standard reference measurement sequence.In the healthy volunteer experiment,five repeated scans without motion and one scan with motion were conducted,and T2 values of seven pairs of regions of interest(ROIs)in different brain areas were extracted.In the motion experiment,images obtained from conventional T2-weighted imaging(T2WI)scans were used as controls.2.Study Population:This study included a total of 79 anterior circulation ischemic stroke patients who underwent scanning using a Siemens Prisma 3.0 T MR scanner,which encompassed conventional MRI scan sequences and MOLED T2 sequence.The severity of motor impairment in all patients was assessed using the upper and lower limb motor function scores extracted from items 5 and 6 of the National Institute of Health Stroke Scale,referred to as the motor index(mNIHSS).Based on the severity of motor impairment,patients were categorized into mNIHSS Ⅰ(mNIHSS score of 0):22 cases,mNIHSS Ⅱ(mNIHSS score of 1～4):45 cases,and mNIHSS Ⅲ(mNIHSS score of 5～8):12 cases.3.Data Processing:The lesion mask comprising ischemic lesions and white matter hyperintensities(WMH)was manually delineated layer by layer on T2 FLAIR images using MRIcron(www.sph.sc.edu/comd/roden/mricron),and the volume of ischemic lesions(ILV)was measured.Registration of the data was performed in FSL(FMRIB Software Library,fsl.fmrib.ox.ac.uk/fsl/fslwiki/Atlases),utilizing the Johns Hopkins University White Matter Tractography Atlas(JHU-CST)as the standard template for the CST.The process involved excluding cerebrospinal fluid(CSF),lesion masks,and CST with probability values less than 20%to obtain the NA-CST in the space of T2 mapping.Subsequently,the average T2 values of the NA-CST on the ipsilesional and contralesional sides were extracted,and the bilateral T2 ratio(rT2)was calculated.4.Statistical analysis:Statistical analysis was performed using SPSS software(version 23.0,IBM Corp,Armonk,NY).To validate the quantitative T2 mapping sequence based on MOLED,Pearson correlation and Bland-Altman analysis were employed,along with tests for normality,coefficient of variation(CoV),intra-group correlation coefficient(ICC)with 95%confidence intervals(CI)to assess accuracy and repeatability.Bland-Altman analysis was used to examine the consistency of scan results with and without motion.When investigating the damage of the NA-CST after stroke,the normal distribution of parameters was first confirmed.Wilcoxon signedrank test was used to compare mean T2 values of NA-CST on the same side as well as the contralateral side of infarcted lesions among all patients and three groups with varying degrees of motor dysfunction.Analysis of variance(ANOVA)and Bonferroni correction were applied for pairwise comparisons of ILV among the three groups with different levels of motor dysfunction.Using Spearman correlation analysis to investigate the correlation between ILV and rT2 values with the severity of motor functional impairment,and the relationship between ILV and mean T2 values of NACST was analyzed.Subsequently,an ordinal Logistic regression model was employed utilizing rT2 and ILV as predictors to estimate the severity of post-stroke motor dysfunction.Finally,the meaningful rT2 value for predicting the severity of motor dysfunction was determined,and the optimal threshold and area under the receiver operating characteristic curve(AUC)were calculated.Results1.Validation Experiment:The validation experiment results demonstrate a strong positive correlation between the quantitative T2 values obtained based on MOLED and MESE,with a Pearson correlation coefficient(PCC)of 0.9995(P<0.0001).Furthermore,the two methods exhibit good consistency,with a mean difference(Meandiff)of 4.20%and a standard deviation of the difference(SDdiff)of 1.71%.Verification experiments with healthy volunteers show excellent repeatability of T2 values obtained based on MOLED in all repeated scans of each volunteer.In the first five motion-free repeated scans,all ROIs demonstrate good consistency,with coefficients of variation(CoV)ranging from 0.47 to 2.74.Additionally,despite motion artifacts affecting conventional sequences during motion experiments,T2 mapping based on MOLED can avoid motion artifacts and acquire high-quality images.BlandAltman analysis results indicate that there is no systematic bias in T2 measurements,regardless of the presence of motion,with a Meandiff of 0.28%and SDdiff of 1.34%.2.Clinical data of ischemic stroke patients:A total of 79 anterior circulation ischemic stroke patients(26 females,53 males)were included in this study,with a mean age of(56.67±11.88)years and an MRI time of 18(8～38)days after stroke onset.Patients were categorized into three groups based on the severity of motor impairment:the mNIHSS I group with 22 cases,the mNIHSS II group with 45 cases,and the mNIHSS III group with 12 cases.There were no statistically significant differences among the three groups of patients in terms of gender distribution,age,and cardiovascular risk factors.3.Comparison of T2 values of the NA-CST on the ipsilesional and contralesional sides after stroke:In all patients,the average T2 value of the infarcted side NA-CST was significantly higher than the contralateral side[T2 ipsilateral:70.05(67.68～73.84);T2 contralateral:68.76(66.13～71.29);P<0.001].Among patients in the mNIHSS I group,there was no significant difference in the average T2 value of the infarcted side and contralateral NA-CST[T2 ipsilateral:68.86(67.27～72.60);T2 contralateral:69.50(67.36～72.96);P=0.485].In patients of the mNIHSS Ⅱ group,the average T2 value of the ipsilateral NA-CST was significantly higher than the contralateral side[T2 ipsilateral:70.60(67.36～73.84);T2 contralateral:68.45(65.55～71.17);P<0.001].For patients in the mNIHSS III group,the average T2 value of the ipsilateral NA-CST was also significantly higher than the contralateral side[T2 ipsilateral:70.75(68.49～80.07);T2 contralateral:69.11(66.58～70.57);P=0.002].4.Comparison of the ILV among three groups of patients with varying degrees of motor dysfunction:There was no significant difference in ILV between patients in the mNIHSS Ⅰ and mNIHSS Ⅱ groups[ILVⅠ:1910(310～3400);ILVⅡ:1640(570～6270);P>0.05].However,significant differences in ILV were observed between patients in the mNIHSS Ⅰ and mNIHSS Ⅲ groups[ILVⅠ:1910(310～3400);ILVⅢ:5740(2250～17780);P=0.014],as well as between patients in the mNIHSS Ⅱ and mNIHSSⅢ groups[ILVⅡ:1640(570～6270);ILVⅢ:5740(2250～17780);P=0.045].5.Correlation analysis between rT2,ILV,and the severity of motor dysfunction:The study results demonstrate a positive correlation between rT2 values and the severity of motor impairment(r=0.575,P<0.001).Higher rT2 values indicate a greater degree of corresponding motor dysfunction.It is worth noting that although there is a correlation between ILV and the severity of motor function(r=0.290,P=0.010),the correlation is relatively weak.Additionally,the study found no correlation between ILV and the average T2 value of the ipsilateral NA-CST(r=0.034,P=0.763).6.Diagnostic efficiency of rT2 on post-ischemic stroke motor function:An ordered Logistic regression analysis using rT2 and ILV as independent variables and the severity of motor dysfunction as the dependent variable revealed a significant predictive value of rT2 for the severity of motor dysfunction.When the critical value of rT2 is set at 1.013,it can effectively determine the occurrence of post-stroke motor impairment,demonstrating a sensitivity of 80.70%and a specificity of 81.82%,with an AUC of 0.883 and an error rate of 0.04.Conclusions1.The MOLED-based quantitative T2 mapping technique possesses certain advantages in image acquisition for patients with ischemic stroke,characterized by its ultra-fast speed,motion robustness,and high accuracy and repeatability.2.The MOLED-based quantitative T2 mapping technique enables the detection of the damage of the NA-CST following ischemic stroke and provides a preliminary assessment of the prognosis of post-ischemic stroke motor impairments.更多还原