基于自动纤维定量法检测肌萎缩侧索硬化症患者脑白质纤维完整性改变的价值  

Value of detecting changes in white matter fiber integrity in patinets with amyotrophic lateral sclerosis based on automatic fiber quantification

作　　者：徐芮 朱思佳 王宁 孔莹 郁义星[1] 蒋彬[1] 万嘉毅 马佳丽 方琪[2] 朱默[1] XU Rui;ZHU Sijia;WANG Ning;KONG Ying;YU Yixing;JIANG Bin;WAN Jiayi;MA Jiali;FANG Qi;ZHU Mo(Department of Radiology,the First Affiliated Hospital of Soochow University,Suzhou 215006,China;Department of Neurology,the First Affiliated Hospital of Soochow University,Suzhou 215006,China)

机构地区：[1]苏州大学附属第一医院放射科,苏州215006 [2]苏州大学附属第一医院神经内科,苏州215006

出　　处：《磁共振成像》2023年第9期44-49,共6页Chinese Journal of Magnetic Resonance Imaging

基　　金：江苏省卫健委老年健康项目(编号:LK2021017)。

摘　　要：目的通过自动纤维定量(automatic fiber quantification,AFQ)技术分析肌萎缩侧索硬化症(amyotrophic lateral sclerosis,ALS)患者脑白质纤维完整性改变,并探究其与支持向量机(support vector machine,SVM)方法联用识别ALS疾病的可行性。材料与方法前瞻性纳入29例ALS患者(ALS组)及相匹配的29例健康对照者(healthy controls,HCs)(HCs组)临床和MRI资料,使用AFQ软件包对所有受试者的扩散张量成像(diffusion tensor imaging,DTI)数据进行分析,追踪全脑20条白质纤维束,每条纤维束分成100等份,定量分析每个等份的各向异性分数(fractional anisotropy,FA)及平均扩散系数(mean diffusivity,MD)、径向扩散系数(radial diffusivity,RD)、轴向扩散系数(axial diffusivity,AD)参数值,运用偏相关分析DTI参数与临床指标的相关性,同时提取两组被试的脑白质纤维差异值作为分类特征,利用SVM区分ALS组和HCs组,估算准确率。结果AFQ结果表明,与HCs相比,ALS患者在左侧皮质脊髓束FA值及AD值降低,左侧额枕下束及右侧上纵束AD值增高,双侧皮质脊髓束MD值、RD值增高,且左侧皮质脊髓束平均FA值与修订版肌萎缩侧索硬化功能评分量表(Amyotrophic Lateral Sclerosis Functional Rating Scale-Revised,ALSFRS-R)中的精细功能域得分呈正相关(r=0.386,P=0.046),右上纵束平均AD值与ALSFRS-R评分中的延髓功能域得分呈正相关(r=0.422,P=0.028),右侧皮质脊髓束平均MD值、RD值均与爱丁堡ALS认知行为量表(Edinburgh Cognitive and Behavioural ALS Screen,ECAS)评分呈负相关(r=-0.428,P=0.026;r=-0.416,P=0.031)。筛选出所有受损纤维束具有组间差异的节点作为特征值,取得了较好的分类效果,对ALS组和HCs组的鉴别准确率达81.00%,且受试者工作特征(receiver operating characteristic,ROC)曲线下面积(area under the curve,AUC)值最大为0.90。结论ALS患者的白质微结构损伤主要与皮质脊髓束有关,基于AFQ分析检测的这些异常可作为一种有效的生�Objective:To analyze the changes of white matter fiber integrity in patients with amyotrophic lateral sclerosis(ALS)by automatic fiber quantification(AFQ)technology,and explore the feasibility of combining it with support vector machine(SVM)to identify ALS disease.Materials and Methods:Clinical and MRI data of 29 patients with ALS(ALS group)and 29 matched healthy controls(HCs group)were prospectively included.Diffusion tensor imaging(DTI)data from all subjects were analyzed using the AFQ software package,and 20 white matter fiber bundles in the whole brain were tracked.Then each fiber bundle was divided into 100 equal parts to acquire quantitative parameter values such as fraction anisotropy(FA),mean diffusion(MD),radial diffusion(RD)and axial diffusion(AD).Partial correlation was further used to explore the relationships between DTI parameters and clinical indicators.Extracting the classification features with white matter fiber difference between two subjects.SVM was used to distinguish them and estimated the accuracy rate.Results:The AFQ results showed that,compared with the HCs,the ALS patients had decreased FA values and AD values in the left corticospinal tract,higher AD values in the left inferior fronto-occipital fasciculus and the right superior longitudinal fasciculus,higher MD and RD values of the bilateral corticospinal tract.The average FA value of the left corticospinal tract was positively correlated with Amyotrophic Lateral Sclerosis Functional Rating Scale-Revised(ALSFRS-R)fine functional domain score(r=0.386,P=0.046)and the average AD value of the right superior longitudinal fasciculus was positively correlated with ALSFRS-R bulbar functional domain score(r=0.422,P=0.028).Both the average MD and RD values of the right corticospinal tract were negatively associated with the Edinburgh Cognitive and Behavioural ALS Screen(ECAS)score(r=-0.428,P=0.026;r=-0.416,P=0.031).All the selected nodes with inter-group differences in damaged fiber tracts were used as feature values to achieve a good classificat

关 键 词：肌萎缩侧索硬化症 扩散张量成像 磁共振成像 自动纤维定量 皮质脊髓束 支持向量机 

分 类 号：R445.2[医药卫生—影像医学与核医学] R746.4[医药卫生—诊断学]