【作者】 覃媛媛；

【导师】 朱文珍；

【作者基本信息】 华中科技大学 ， 影像医学与核医学， 2013， 博士

【摘要】 第一部分APP/PS1转基因小鼠活体全脑DTI定量研究目的：以往的研究已将扩散张量成像(diffusion tensor imaging, DTI)应用于阿尔茨海默病(Alzheimer’s disease, AD)动物模型的组织病理学研究中,但是少有关于结构特异性方面的报道。基于体素的分析方法(voxel-based analysis, VBA)和基于解剖图谱的分析方法(atlas-based analysis, ABA)是DTI全脑分析方法中两种互补的方法。本研究的目的在于采用全脑DTI的分析方法,明确AD动物模型病理变化的空间位置分布特征。材料与方法：同时采用VBA和ABA的方法,对APP/PS1转基因小鼠(n=9)和野生型对照(n=9)进行全脑的DTI对比分析。采用多种度量指标,如各向异性分数(fractional anisotropy, FA)、扩散轨迹(total diffusivity, trace)、轴向弥散(axial diffusivity, DA)和放射弥散(radial diffusivity, DR)对阿尔茨海默病小鼠不同类型的病理变化进行量化分析。采用Kappa分析的方法对手动描绘的感兴趣区(region of interest, ROI)和基于解剖图谱方法所勾画的ROI进行比较,以评估图像配准的准确性。MR检查之后,对APP/PS1转基因小鼠和野生型对照进行组织学检查分析。结果：结果显示,APP/PS1转基因小鼠存在广泛的脑结构异常,包括灰质区域如新皮层、海马、纹状体、丘脑、下丘脑、屏状核、杏仁核及梨状皮层,和白质区域如胼胝体/外囊、扣带束、隔、内囊、海马伞及视束,均表现为FA值或DA值升高,或者FA值和DA值同时升高(p<0.05,FDR校正)。手动描绘的ROI与ABA方法所描绘的ROI之间的平均Kappa值均接近0.8,且在APP/PS1转基因小鼠组和野生型对照组之间无显著性差异(p>0.05)。组织病理学分析证实了灰质区域如新皮层和海马区微结构的DTI变化。DTI同时也发现了广泛的白质区域的弥散改变,但这种差异仅靠单层的组织学定性观察难以准确评估。结论：本研究报道了APP/PS1转基因小鼠脑结构特异性的病理变化,同时也证实了全脑DTI定量分析方法在AD动物模型中的可行性。第二部分AD、MCI和健康人群脑白质差异的空间分布模式探讨目的：近年来大量研究均发现阿尔茨海默病(AD)患者、轻度认知障碍(MCI)患者和健康人群的脑白质完整性存在显著差异,然而AD和MCI患者脑白质损害的空间分布模式少有报道。本研究旨在通过全脑的DTI定量分析,探讨AD、MCI和健康人群脑白质差异的空间分布模式,找到疾病早期诊断和监测疾病进展的可靠指标。材料与方法：依据NINCDS-ADRDA可能AD的标准纳入AD患者21例(M/F=11/10,平均年龄66.8岁)；依据Petersen的标准纳入MCI患者8例(M/F=3/5,平均年龄64.4岁)；及无神经系统疾病的健康对照15例(M/F=8/7,平均年龄65.3岁)。采用GE公司signa HDxt3.0Tesla超导磁共振扫描仪行扩散张量成像(diffusion tensor imaging,DTI),扫描参数如下：TR/TE=10000/83ms, FA=90°, Matrix=256x256, FOV=240mmx240mm, Phase FOV=1,层厚3.0mm无间隔,NEX=1,42层覆盖全脑,b值为1000s/mm2,30个方向。得到DTI原始图像之后,利用DTIstudio软件进行FA图重建,利用DiffeoMap软件对图像进行基于解剖图谱的分析,测量深部灰质和深部白质共58个脑区结构的FA值。AD、MCI和健康对照组58个脑区结构的FA值首先采用单因素方差分析并进行事后检验,两两比较组间差异；然后对相关脑区FA值与简易精神状态量表(mini-mental state examination, MMSE)评分做相关分析。结果：与健康人群相比,AD患者深部灰质和深部白质结构存在广泛的FA值降低(p<0.05,FDR校正)。其中,胼胝体压部和丘脑的FA值在MCI组和健康对照组间存在显著差异(p<0.05,FDR校正),但在AD组和MCI组间无差异(p>0.05)；扣带束和上纵束等8个结构的FA值在AD组和MCI组间有显著差异(p<0.05,FDR校正),但在MCI组和健康对照组间无差异(p>0.05)。相关分析显示,扣带束和上纵束的FA值与MMSE评分存在显著的正相关关系,以右侧扣带束的相关系数值最高(r=0.606,p=0.001)；而胼胝体压部和丘脑区域FA值与MMSE不存在相关关系(p>0.05)。结论：AD和MCI患者脑白质损害的空间分布模式存在显著差异。胼胝体压部和丘脑显微结构病变是早期事件,与认知功能下降关系不大。而扣带束和上纵束白质病变与疾病进展有关,与认知功能下降显著相关。第三部分定量结构MRI对阿尔茨海默病的鉴别诊断研究目的：提出一种全新的方法,可将脑部T1加权磁共振(magnetic resonance, MR)图像转变为特征矢量,应用于基于内容的图像检索(content-based image retrieval, CBIR)。为了克服临床中同一人群的解剖学个体差异及成像参数的不一致性,我们提出了一种基于目标图像与解剖图谱之间差异的图像分析方法(Gap between an Atlas and a target Image Analysis, GAIA),利用基于解剖图谱的图像分割方法(atlas-based analysis, ABA),寻找目标图像与解剖图谱之间差异的大小,从中提取目标图像的解剖学特征,用于阿尔茨海默病的鉴别诊断研究。材料与方法：选取阿尔茨海默病(Alzheimer’s disease, AD)、亨廷顿病(Huntington’s disease, HD)、脊髓小脑性共济失调6型(Spinocerebral ataxia type6, SCA6)、原发性进行性失语症(primary progressive aphasia, PPA)患者及正常人的T1加权MR图像共102例,作为训练数据。另外随机选取AD、HD、SCA6、PPA患者及正常人的T1加权MR图像共170例作为测试数据。采用GAIA的方法对训练数据进行模式分类,分别提取AD、HD、SCA6、PPA患者及正常人的神经解剖学特征作为特征矢量；随后将这些特征矢量应用到测试数据中,每一个测试数据分别得到一个判别得分(discriminant score),利用判别得分对其进行病种的判别,并评估GAIA判别不同种类疾病的准确性。结果：从训练数据中提取出来的特征矢量,与我们所选取的各神经变性疾病所对应的病理学标志完全一致。大部分测试数据的判别得分能够准确的将其分类至各自对应的疾病种类中去。不具备该疾病典型相关解剖学特征的数据不能被准确分类。GAIA可将阿尔茨海默病从其它类型的神经变性疾病中区分开来。结论：我们提出的GAIA方法,是基于疾病相关的解剖学特征的提取方法,在图像的特征提取与模式识别中有着广阔的应用前景。在未来,可使得放射科医生只需要提交一名患者的图像,就能够将具有类似解剖学特征的相关临床病例全部检索出来,从而对某种疾病的诊断、治疗、预后及随访预测进行大样本的人口学普查及统计分析。更多还原

【Abstract】 Part I In vivo Quantitative whole-brain Diffusion Tensor Imaging analysis of APP/PS1transgenic mice using Voxel-based and Atlas-based methodsPurpose Diffusion tensor imaging (DTI) has been applied to characterize the pathological features of Alzheimer’s disease (AD) in a mouse model, although little is known about the structural specificity. Voxel-based analysis (VBA) and atlas-based analysis (ABA) are good complementary tools for whole-brain DTI analysis. The purpose of this study is to identify the spatial localization of disease-related pathology of AD mouse model.Materials and Methods VBA and ABA quantification were used for the whole-brain DTI analysis of nine APP/PS1mice and wild-type (WT) controls. Multiple scalar measurements, including fractional anisotropy (FA), trace, axial diffusivity (DA), and radial diffusivity (DR), were investigated to capture the various types of pathology. The accuracy of the image transformation applied for VBA and ABA was evaluated by comparing manual and atlas-based structure delineation using kappa statistics. Following the MR examination the brains of the animals were analyzed for microscopy.Results Extensive anatomical alterations in APP/PS1mice, including in both gray matter areas (neocortex, hippocampus, caudate putamen, thalamus, hypothalamus, claustrum, amygdala, and piriform cortex) and white matter areas (corpus callosum/external capsule, cingulum, septum, internal capsule, fimbria, and optic tract) have been identified by an increase in FA or DA, or both, compared to WT (p<0.05, corrected). The average kappa value between manual and atlas-based structure delineation was approximately0.8, and there was no significant difference between APP/PS1and WT mice (p>0.05). The histopathological changes in the gray matter areas were confirmed by microscopy studies. DTI did, however, demonstrate significant changes in white matter areas, where the difference was not apparent by qualitative observation of a single-slice histological specimen.Conclusion This study demonstrated the structure specificity of pathological changes in APP/PS1mouse model, and also showed the feasibility of applying whole-brain analysis methods for the investigation of AD mouse model. Part II The Pattern of White Matter changes among Alzheimer’s Disease, Mild Cognitive Impairment and Healthy peoplePurpose Increasing evidence has demonstrated that white matter(WM) changes among Alzheimer’s disease (AD), mild cognitive impairment (MCI) and healthy people are signficantly different. However, the pattern of WM changes are still under debate. The purpose of this study is to identify the spatial pattern of WM alterations among AD, MCI and healthy people, and to find reliable biomarkers for the early diagnosis and monitoring of the disease.Materials and Methods Twenty-one patients diagnosed as probable AD according to NINCDS-ADRDA(M/F=11/10, mean age66.8yrs.),8patients diagnosed as MCI according to Petersen’s criteria (M/F=3/5, mean age64.4yrs.) and15healgy people (M/F=8/7, mean age65.3yrs.) were enrolled in this study. All subjects underwent diffusion tensor imaging (DTI) on a3.0T MR system with TR/TE of10000/83ms, FA of90°, matrix of256×256, FOV of240mm×240mm, Phase FOV of1, slice thickness of3.0mm with no space, NEX of1, total slice of42, b value of1000s/mm2along30directions. All the raw data was processed by using DTI studio software to get the fractional anisotropy (FA) images. Then atlas-based analysis (ABA) were used for whole-brain DTI anlysis including58deep gray matter (GM) and deep WM structures. The differences of FA value among AD, MCI and healthy people were compared by using ANOVA, with a post-hoc analysis. The correlation between FA value and MMSE scores were further investigated in the regions where significant differences were found. Results Compared with healthy controls, AD patients demonstrated wide-spread FA decrease in deep GM and deep WM structures (p<0.05, FDR corrected). Among all the structures, the FA value of the splenium of corpus callosum (SCC) and the thalamus were signficantly different between the MCI group and the healthy group (p<0.05, FDR corrected), but not between the AD group and the MCI group (p>0.05); the FA value of the cingulum and the superior longitudinal fasciculus (SLF) were significantly different between the AD group and the MCI group (p<0.05, FDR corrected), but not between the MCI group and the healthy group (p>0.05). The mean FA value of the cingulum and the SLF were positively correlated with MMSE scores, with the highest correlation coefficient in the right cingulum (r=0.606, p=0.001). No significant correlation was found between the FA value of SCC and MMSE score (p>0.05), or the thalamus and MMSE score (p>0.05).Conclusion The spatial pattern of WM alterations among AD, MCI and healthy people are significantly different. The microstructure changes in the SCC and the thalamus are early events, but have no significant correaltion with the cognition impairment. The WM disruption in the cingulum and the SLF are in correlation with cognition decline, suggesting that FA values in these areas could be used as a sentive biomarker for monitoring disease progression. Part Ⅲ Gap between an Atlas and a Target Image Analysis (GAIA):Use of a Degree of Local Atlas-Image Segmentation Disagreement to Capture the Features of Anatomic Brain MRIPurpose To develop a new method to convert T1-weighted brain MRIs to feature vectors, which could be used for content-based image retrieval (CBIR). To overcome the wide range of anatomical variability in clinical cases and the inconsistency of imaging protocols, we introduced the Gap between an Atlas and a target Image Analysis (GAIA), in which a degree of local atlas-image segmentation disagreement was used to capture the anatomical features of target images.Materials and Methods As a proof-of-concept, the GAIA was applied to a training dataset for pattern recognition of the neuroanatomical features of Alzheimer’s disease, Huntington’s disease, spinocerebellar ataxia type6, and four subtypes of primary progressive aphasia. These feature vectors were applied to the test dataset to evaluate the accuracy of the pattern recognition.Results The feature vectors extracted from the training dataset agreed well with the known pathological hallmarks of the selected neurodegenerative diseases. Overall, discriminant scores of the test images accurately categorized these test images to the correct disease categories. Images without typical disease-related anatomical features were misclassified.Conclusion The proposed method is a promising method for image feature extraction based on disease-related anatomical features, which will enable users to submit a patient image and search past clinical cases with similar anatomical phenotypes.更多还原