基于BOLD-fMRI的脑胶质瘤分级及代偿机制的研究

发布时间：2018-04-30 23:17

  本文选题：基于血氧水平依赖的功能磁共振成像 + 胶质瘤　； 参考：《南京航空航天大学》2014年硕士论文

【摘要】：神经胶质瘤是大脑中最为常见且危害最为严重的肿瘤病症,临床缺乏有效的根治方法。神经胶质瘤具有弥漫性生长的特性,对认知功能的影响往往随着占位位置的改变而存在差异性,给临床医生制定有效的手术计划,同时最大程度地保护认知功能带来挑战。基于血氧水平依赖的功能磁共振成像(blood oxygenation level-dependent functional magnetic resonance imaging,BOLD-f MRI)技术具有无创、高时空分辨率等多重优势,已成为脑功能研究的重要工具。而静息态BOLD-f MRI(resting state blood oxygenation level-dependent functional magnetic resonance imaging,RS-f MRI)不需要复杂精细的实验设计,也不需要患者进行任务配合,更易被医生和患者接受,也更适合临床上的研究与应用。本文通过BODL-f MRI对胶质瘤的分级和功能代偿进行系统研究,为胶质瘤诊断和治疗提供更多参考信息,便于制定更为合理的临床手术计划。论文研究内容及主要创新点如下:1.基于支持向量机(support vector machine,SVM)的胶质瘤分级研究。首先基于RS-f MRI构建胶质瘤特征参数,包括信号强度差异比(signal intensity difference ratio,SIDR)、信号强度相关性(signal intensity correlation,SIC)、低频振幅分数(fractional amplitude of low-frequency fluctuation,f ALFF)和局域一致性(regional homogeneity,Re Ho),然后利用Mann-Whitney U检验对各参数与高低级别胶质瘤之间的差异性进行检验,最后通过SVM对各个参数的分级准确性、敏感性和特异性进行分析。研究结果表明,SIC和f ALFF参数在高低级别之间具有显著性差异,SIC具有最优的SVM分类效果,分类准确率为89%,其他各个参数的分类准确性、敏感性和特异性都超过了50%。2.基于静息态脑功能网络(resting state network,RSN)的额叶胶质瘤病人代偿机制研究。首先基于左额叶胶质瘤病人、右额叶胶质瘤病人和健康对照组的RS-f MRI数据,分别构建出大脑功能网络,并计算出各个网络的全脑、左脑和右脑的节点度(K)和介数中心度(BC)参数;其次对各个网络的核心节点进行比较,分析各个网络的核心节点差异;最后将病人的网络参数与MMSE评分进行相关性分析。研究结果表明,额叶肿瘤患者病灶侧的K值和BC值始终低于对侧正常区域,同时大脑的核心节点也呈现出向对侧正常区域转移的趋势;从MMSE评分与大脑网络参数的相关性分析发现,认知功能与脑网络参数存在显著相关性。
[Abstract]:Glioma is the most common and serious tumor disease in the brain. Glioma has the characteristics of diffuse growth, and the influence on cognitive function is often different with the change of location, which brings challenges to clinicians in making effective operation plan and protecting cognitive function to the greatest extent. Blood oxygenation level-dependent functional magnetic resonance imaging based on blood oxygen level dependence has many advantages, such as noninvasive, high spatial and temporal resolution, and has become an important tool for the study of brain function. The rest BOLD-f MRI(resting state blood oxygenation level-dependent functional magnetic resonance imagingRS-f MRI does not require complicated and precise experimental design, nor does it need patients to cooperate with the task. It is more easily accepted by doctors and patients, and is more suitable for clinical research and application. In this paper, the grading and functional compensation of gliomas are systematically studied by BODL-f MRI, which provides more reference information for the diagnosis and treatment of gliomas and facilitates the establishment of a more reasonable clinical operation plan. The main contents and innovations of this paper are as follows: 1. A study of glioma classification based on support vector machine (SVM) vector machine machine (SVM). Firstly, the characteristic parameters of glioma were constructed based on RS-f MRI. These include signal intensity difference ratio, signal intensity correlation, fractional amplitude of low-frequency configuration of low frequency and local consistency. Then the differences between parameters and grades of gliomas are examined by Mann-Whitney U test. Finally, the classification accuracy, sensitivity and specificity of each parameter were analyzed by SVM. The results show that there are significant differences between the parameters of sic and f ALFF. Sic has the best effect of SVM classification, and the accuracy of classification is 89. The accuracy, sensitivity and specificity of the other parameters are more than 50.2. Study on compensatory Mechanism of frontal Glioma patients based on resting state Network. Firstly, based on the RS-f MRI data of patients with left frontal glioma, patients with right frontal gliomas and healthy controls, the brain functional networks were constructed, and the parameters of the whole brain, the nodal degree of the left brain and the right brain were calculated. Secondly, the core nodes of each network are compared, and the differences between the core nodes of each network are analyzed. Finally, the correlation between the patient's network parameters and the MMSE score is analyzed. The results showed that the K and BC values of the lesion side of frontal lobe tumor patients were always lower than those of the contralateral normal region, and the core nodes of the brain also showed a tendency of metastasis to the contralateral normal region. From the correlation analysis between MMSE scores and brain network parameters, it was found that cognitive function was significantly correlated with brain network parameters.
【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R739.41
，

本文编号：1826671