基于统计参数图方法对颞叶癫痫PET定侧诊断价值的研究

发布时间：2018-07-01 20:08

本文选题：正电子 + 发射型　；参考：《第二军医大学》2016年博士论文

【摘要】：研究背景:癫痫是以一种由于神经元兴奋与抑制失衡并导致神经元群同步异常放电为特征的中枢神经系统功能失常综合征。据流行病学统计,我国癫痫患病率约为3.6~7.0‰,在神经系统病变中仅次于脑卒中为第二大常见疾病,其中颞叶癫痫(Temporal Lobe Epilepsy,TLE)约占60~70%。在大部分患者中,药物治疗可使得症状得以有效控制,但也有将近1/3的患者会发展为药物难治性癫痫,其中,颞叶内侧硬化患者常为药物难治性癫痫且对手术治疗响应较好、需要手术治疗,手术治疗主要以对癫痫发作区域的切除为主。对药物难治性癫痫来讲,完善的术前评估是保证其疗效的基础,但常受到难以准确勾画致痫灶的困扰。目前来看,多种功能分子成像技术可能在判定致痫灶定侧、定位方面提供一些补充信息,其中正电子发射型计算机断层扫描(Positron Emission Computed Tomography,PET)脑显像技术是一种无创性的、能够提供定位、定量信息的脑功能代谢检查技术,在临床已有较多应用。既往PET脑显像评价主要依靠视觉评价以及脑内不对称指数计算等方法,虽取得较好临床诊断效果,但其主观影响大,受诊断医师水平影响较大,亟须一种更加客观、准确的分析方法,统计参数图在颞叶癫痫上进行分析既往国内外有一定研究,但因研究对象及对照组例数较少、多方法综合比较研究较少等原因,其诊断价值尚不确切、肯定,未能真正进入临床应用。研究目的:本研究旨在利用统计参数图(Statistical Parametric Mapping,SPM)方法对颞叶癫痫患者定侧诊断进行研究、开发,以患者术后病理及随访结果为金标准,对不同方式SPM处理方法诊断效能进行比较分析,并将所采用的SPM方法分与临床常规应用的视觉分析方法、不对称指数(Asymmetry Index,AI)分析法等所得结果进行统计学比较分析,评价各方法诊断效能,选定并建立合适的SPM方法学、优选相关参数,在此基础上评价SPM分析法的价值,此外,通过患者组与对照组间的组间比较,初步讨论颞叶癫痫时的机制问题。其研究意义在于建立适当、准确的基于像素水平的统计参数图对颞叶癫痫进行定侧诊断的方法,取得分析时应采用的各项合理参数,对SPM方法在颞叶癫痫中的诊断价值进行评价。研究方法:对86例资料完整的难治性颞叶癫痫患者及37例健康志愿者进行回顾性分析,分析方法包括三大部分:(1)、视觉分析方法,包括独立双盲目测法和视觉AI分析法阅片,记录研究对象双侧颞叶SUVmean值并将患者组及健康对照组双侧颞叶SUVmean值进行比较,将患者组患侧SUVmean值与健康对照组相应值进行比较,将目测法和视觉AI分析法阅片所得准确率进行比较。此外,我们利用两组研究对象的SUVmean值进行ROC曲线分析。(2)、利用SPM分析方法对难治性颞叶癫痫进行定侧诊断研究,采用不同置信水平P值分别进行两方面研究,(1)将患者组的每个患者数据与健康对照组进行基于像素水平两样本t检验;(2)与对照组内存在的脑内不对称进行比较,并藉此评价每个患侧脑内的不对称情况。每个患者图像及对照组沿着矢状位进行左右翻转得到镜像图像,将患者组、对照组原始图像及两组反转镜像图像进行单因素方差分析;(3)分别对不同SPM方法诊断准确率进行分析,将两组所得最优置信水平的方法进行比较,其后将最优化SPM诊断方法与视觉分析方法进行比较,评价相应SPM诊断方法的诊断效能。(3)、将左侧颞叶癫痫患者组、右侧颞叶癫痫患者组及全部癫痫患者组分别与对照组进行组间比较,评价前述各患者组与对照组间脑区代谢差异。结果:(1)目测方法准确检出70例患者致痫灶,视觉AI分析法中72例患者能够明确定侧诊断,两组间诊断准确率无统计学显著差异;经统计学分析,健康对照组双侧颞叶SUVmean比较无统计学显著差异;TLE组患侧SUVmean低于健侧相应值,有统计学显著差异;TLE组患侧SUVmean明显低于健康对照组左、右侧颞叶SUVmean,有统计学显著意义;利用SUVmean作为致痫灶定侧指标行ROC曲线分析得:当SUV值为4.35时,其对癫痫灶定侧诊断的敏感度为87.5%,特异性为81.4%,曲线下面积0.844。(2)采用不同置信水平P值,将单个患者与对照组进行两样本t检验,发现当P0.05(FWE校正值)时诊断准确率为79.07%(68/86),优于其它置信水平,与视觉分析法相当;采用不同置信水平P值,利用SPM计算单个患者脑内不对称性,发现当选取置信水平P0.05时诊断准确率可达94.19%(81/86),优于采用其它种置信水平所得结果,优于视觉分析方法及单个患者与对照组进行两样本t检验方法。(3)组分析研究发现,除颞叶代谢减低外,其他部分脑区有代谢减低或增高改变,可部分反映颞叶癫痫发病机制,并有助于临床医生认识癫痫典型影像变化特点。结论:总体来看,SPM是通过对脑内所有象素进行逐一进行分析、比较而最终形成统计参数图,故其对病变的探测级别可以达到象素水平,SPM脑内不对称分析的PET分析方法基于像素水平进行统计分析,考虑了部分客观因素,并排除了诊断医师主观因素等影响,可无创地、更加客观、科学、全面、准确地发现各脑区微小代谢变化,对致痫灶检出率高、灵敏度、特异性强、可重复性强,在难治性TLE定侧诊断上有较高的临床应用价值,因其流程化操作特点,有望进入临床实践。
[Abstract]:Background: epilepsy is a central nervous system dysfunction syndrome characterized by the imbalance of neuronal excitability and inhibition and the synchronous abnormal discharge of the neurons. According to epidemiological statistics, the prevalence of epilepsy in China is about 3.6~7.0 per thousand. In the pathological changes of the nervous system, only stroke is second common diseases, of which the temporal lobe is in the temporal lobe. Temporal Lobe Epilepsy (TLE) accounts for about 60~70%. in most of the patients. Drug treatment can effectively control the symptoms, but there are also nearly 1/3 patients who will develop into drug refractory epilepsy. Among them, the patient with Chang Wei's medial temporal sclerosis is difficult to treat and responds well to the surgical treatment. Surgical treatment is needed and the operation is the main treatment. In the case of drug refractory epilepsy, a perfect preoperative assessment is the basis for the treatment of epilepsy, but it is often difficult to accurately describe the preeclampsia. At present, multiple functional molecular imaging techniques may provide some additional information for the determination of the localization of the epileptic foci and the location of the positron. Positron Emission Computed Tomography (PET) brain imaging technology is a noninvasive brain functional metabolic examination technique that provides location and quantitative information. It has been widely used in clinical practice. The evaluation of PET brain imaging is mainly based on visual evaluation and the calculation of asymmetric index in the brain. But it has a better clinical diagnosis effect, but its subjective influence is big and the diagnostic physician level is greatly influenced. It is urgent to have a more objective and accurate analysis method. The statistical parameter map has a certain research on the temporal lobe epilepsy, but the research object and the control group are few, and the multiple methods are comparatively few and so on. The purpose of this study is to study the lateral diagnosis of temporal lobe epilepsy by using the Statistical Parametric Mapping (SPM) method. The purpose of this study is to develop the diagnostic efficiency of different methods of SPM treatment by using the postoperative pathological and follow-up results as the gold standard. The comparative analysis was carried out, and the SPM method was compared with the visual analysis method of clinical routine application, the results of asymmetric index (Asymmetry Index, AI) analysis were compared, the diagnostic efficiency of each method was evaluated, and the appropriate SPM methodology was established and the relevant parameters were selected. On this basis, the price of the SPM analysis method was evaluated. In addition, the mechanism of temporal lobe epilepsy was preliminarily discussed by comparison between the patient group and the control group. The significance of this study was to establish a proper and accurate method for the lateral diagnosis of temporal lobe epilepsy based on the pixel level statistical parameter map, and to obtain various reasonable parameters for the analysis of temporal lobe epilepsy, and the SPM method was used in temporal lobe epilepsy. The diagnostic value of 86 cases of intractable temporal lobe epilepsy and 37 healthy volunteers were analyzed retrospectively. The analysis method included three parts: (1) the visual analysis method, including the independent double blind method and the visual AI analysis, to record the SUVmean value of the bilateral temporal lobe and the patient group. The SUVmean values of bilateral temporal lobes in the healthy control group were compared and compared with the corresponding values of the patients' affected side SUVmean and the healthy control group. The accuracy rate of the visual and visual AI analysis was compared. In addition, we used the SUVmean value of the two groups of subjects to analyze the ROC curve. (2) the treatment of intractable temporomandibular temporomandibular time was made by SPM analysis. Ye Dianxian carried out the study of lateral diagnosis with two different confidence level P values. (1) the data of each patient in the patient group was compared with the healthy control group based on the pixel level two sample t test, and (2) compared with the intracerebral asymmetry in the control group, and to evaluate each side of the affected side of the brain. The patient image and the control group flipped along the sagittal position to get the image image. The patient group, the original image of the control group and the two groups of reverse mirror images were analyzed by single factor variance analysis. (3) the diagnostic accuracy of different SPM methods was analyzed, and the optimal confidence level of the two groups was compared, and then the optimal SPM diagnosis was taken. Compared with the visual analysis method, the diagnostic efficiency of the corresponding SPM diagnosis method was evaluated. (3) the left temporal lobe epilepsy group, the right temporal lobe epilepsy group and all the epileptic patients were compared with the control group respectively, and the metabolic difference between the foregoing group and the control group was evaluated. Results: (1) the visual inspection method was accurate. 70 cases of patients with epileptic foci, 72 cases of visual AI analysis can be definite side diagnosis, the diagnostic accuracy rate between the two groups is not statistically significant difference. After statistical analysis, there is no statistically significant difference in the bilateral temporal lobe SUVmean in the healthy control group; the affected side of the TLE group is lower than the corresponding value of the healthy side, there is a statistically significant difference; the TLE group suffers from the side SUVmean. The left side of the right temporal lobe was significantly lower than that in the healthy control group. The SUVmean was statistically significant in the right temporal lobe, and the ROC curve was analyzed by SUVmean as an index of the epileptic focus side. When the SUV value was 4.35, the sensitivity to the localized diagnosis of epileptic foci was 87.5%, the specificity was 81.4%, and the area under the curve was 0.844. (2) using the P values of different confidence levels. The two sample t test showed that the diagnostic accuracy of the P0.05 (FWE correction value) was 79.07% (68/86), which was superior to the other confidence levels, and was equivalent to the visual analysis method. Using the P values of different confidence levels and using SPM to calculate the asymmetry in the brain of a single patient, it was found that the diagnostic accuracy of the elected level P0.05 was 94.19% (81/86), which was better than that adopted. The results obtained from other levels of confidence were better than the visual analysis method and the two sample t test for individual patients and the control group. (3) the analysis of group analysis found that the metabolism decreased or increased in other parts of the brain except the decrease of temporal lobe metabolism, which could partly reflect the mechanism of temporal lobe epilepsy and help clinicians to recognize typical images of epilepsy. Like change characteristics. Conclusion: in general, SPM is by analyzing all the pixels in the brain, comparing and finally forming a statistical parameter map, so the detection level of the disease can reach the level of pixels. The PET analysis method of SPM asymmetry analysis in the brain is based on the statistical analysis of pixel leveling, considering some objective factors and side-by-side. In addition to the influence of the doctor's subjective factors, it can be noninvasive, more objective, scientific, comprehensive and accurate to find small metabolic changes in the brain regions, high detection rate, sensitivity, specificity and repeatability of the epileptic foci. It has high clinical application value in the diagnosis of intractable TLE, and it is expected to enter clinical reality because of its flow operation characteristics. Practice.
【学位授予单位】：第二军医大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：R742.1;R817.4

【相似文献】