垂体腺瘤侵袭性的DCE-MRI参数直方图分析及其与免疫组化关联性研究

发布时间：2018-04-27 10:33

本文选题：侵袭性垂体腺瘤 + DCE-MRI　；参考：《大连医科大学》2017年硕士论文

【摘要】：目的:应用DCE-MRI参数(Ktrans值、Ve值、Kep值)直方图对侵袭性与非侵袭性垂体腺瘤进行全域定量分析,并探究其与Ki-67、p53之间的相关性,评估DCE-MRI参数直方图对垂体腺瘤侵袭性的诊断价值。材料与方法:收集经手术及病理证实的37例垂体腺瘤患者的临床资料、术前MRI资料、术后病理和p53、Ki-67资料,按Knosp分级分为侵袭组(18例)与非侵袭性组(19例)。MRI扫描采用美国GE SignaHDXT3.0T MR机及配套的8通道头颅线圈检查,扫描序列包括垂体常规平扫和动态增强(Dynamic Contrast Enhanced MRI,DCE-MRI),DCE-MRI扫描采用LAVA-Flex序列,连续扫描30期。将DCE-MRI的DICOM格式原始数据拷贝至个人电脑,使用Omni-Kinetics软件进行后处理。测量肿瘤最大径;参照冠状位T1WI图像,在包含肿瘤的每一个层面上沿肿瘤边缘手动描绘ROI,当肿瘤包绕颈内动脉时避开颈内动脉。将所有层面的ROI累加为一个3D ROI,软件将自动计算出Ktrans值、Ve值、Kep值的直方图及其所有参数,包括:最小值、最大值、平均值、第5百分位数、第10百分位数、第25百分位数、第50百分位数、第75百分位数、第90百分位数、第95百分位数、值域、标准差、方差、平均差、相对差、偏度、峰度、一致性、能量值及熵值。应用社会科学统计软件包(statistics package for social science,SPSS)18.0 版进行数据分析:当数据符合正态分布且方差齐性时,以"均数士标准差"表示,应用单因素方差分析与两独立样本t检验,不符合正态分布者以"中位数士四分位间距"表示,采用Kruskal-Wallis检验与Mann-Whitney U检验,分析侵袭组及非侵袭组肿瘤的最大径、Knosp分级、Ki-67、p53及Ktrans值、Ve值、Kep值直方图参数的差异性;应用Spearman相关性分析knosp分级与ki-67、p53之间的相关性,Ktrans值、Ve值、Kep值直方图参数与垂体腺瘤knosp分级、与ki-67、p53之间的相关性;利用接受者操作特性(receiver operating characteristic,ROC)曲线来确定肿瘤最大直径、Ki-67、p53、Ktrans值、Ve值、Kep值直方图参数对二者的诊断能力;应用Logistic回归分析模型得到的联合变量,并利用ROC曲线来确定对二者的诊断能力。p0.05为差异有统计学意义。结果:1.侵袭性与非侵袭性垂体腺瘤Ktrans值直方图参数中最大值、平均值、第50百分位数、第75百分位数、第90百分位数、第95百分位数、值域、平均差、偏度、一致性、能量值、熵值在两组间有显著差异(P=0.016,0.002,0.025,0.010,0.006,0.003,0.012,0.008,0.025,0.012,0.028,0.024);利用平均数、第 95百分位数、第90百分位数鉴别侵袭性与非侵袭性垂体腺瘤ROC曲线的效能较好,AUC 分别为 0.792、0.784、0.766;2.侵袭性与非侵袭性垂体腺瘤Ve值直方图参数中最大值、平均值、第95百分位数、值域、标准差、方差、相对差、偏度、峰度、一致性、能量值、熵值在两组间有显著差异(P=0.006,0.036,0.023,0.013,0.029,0.027,0.005,0.042,0.013,0.022,0.033,0.046);利用相对差、最大值、值域鉴别二者的效能较好,AUC 分别为 0.772、0.766、0.741;3.侵袭性与非侵袭性垂体腺瘤Kep值直方图参数中第90百分位数、第95百分位数、值域、相对差、偏度在两组间有显著差异(P=0.031,0.013,0.042,0.005,0.023);利用相对差、偏度、第95百分位数鉴别二者的效能较好,AUC分别为0.769、0.740、0.737;4.Ktrans值平均数、Ve值相对差、Kep值相对差融合成的联合变量1,及联合变量1与肿瘤最大径融合成的联合变量2,通过ROC曲线鉴别侵袭性与非侵袭性垂体腺瘤的AUC分别为0.877、0.942;5.垂体腺瘤knosp分级与ki-67、p53之间具有较好的相关性(r=0.547,P0.001;r=0.617,P0.001);6.垂体腺瘤DCE-MRI参数直方图中Ktrans值直方图平均值与Knosp分级的相关性最佳(r=0.660,p0.001),Kep值直方图峰度与Ki-67的相关性最佳(r=0.746,p0.001),Ktrans值直方图平均差与p53的相关性最佳(r=0.388,p=0.018)。结论:1.DCE-MRI参数直方图可反映侵袭性垂体腺瘤微血管异质性,有助于侵袭性垂体腺瘤和非侵袭性垂体腺瘤的鉴别;2.联合应用DCE-MRI参数直方图及肿瘤最大径可提高对垂体腺瘤侵袭性的诊断;3.垂体腺瘤Kep值直方图峰度、Ktrans值直方图平均差分别与Ki-67、p53有.一定相关性,可以作为监测病情演变的影像学指标。
[Abstract]:Objective: to use DCE-MRI parameters (Ktrans, Ve, Kep) histogram for quantitative analysis of invasive and non-invasive pituitary adenomas, and to explore the correlation between Ki-67 and p53, and to evaluate the diagnostic value of DCE-MRI parameter histogram for pituitary adenoma invasiveness. Materials and methods: 37 cases of pituitary adenoma confirmed by surgery and pathology were collected. The patients' clinical data, preoperative MRI data, postoperative pathology and p53, Ki-67 data were divided into invasive group (18 cases) and non invasive group (19 cases) with.MRI scan of American GE SignaHDXT3.0T MR machine and 8 channel head coils. The scan sequence included routine plain scan and dynamic enhancement (Dynamic Contrast Enhanced MRI). E-MRI), DCE-MRI scan uses LAVA-Flex sequence and continuously scan 30 phases. Copy the DICOM original data of DCE-MRI to personal computer, use Omni-Kinetics software for post processing. Measure the maximum diameter of tumor; refer to the coronary T1WI image, manually depict ROI on the edge of the tumor at each level of the tumor and move around the neck around the neck. Pulse to avoid the internal carotid artery. Add all levels of ROI into a 3D ROI, and the software will automatically calculate the Ktrans, Ve, Kep value histogram and all the parameters, including the minimum, maximum, average, fifth percentile, tenth percentile, twenty-fifth percentile, fiftieth percentile, seventy-fifth percentile, ninetieth percentile, ninety-fifth 100 Quantile, range, standard deviation, variance, mean deviation, mean deviation, relative deviation, deviation, kurtosis, consistency, energy value and entropy. Data analysis is carried out by the 18 edition of statistics package for social science, SPSS): when the data conforms to normal distribution and the variance is homogeneous, the single factor variance is applied to the single factor variance. The analysis and two independent sample t test did not conform to the normal distribution with the "median interval of four points". Kruskal-Wallis test and Mann-Whitney U test were used to analyze the maximum diameter, Knosp classification, Ki-67, p53 and Ktrans value, the difference of the Ve value and Kep value histogram parameters of the invasive and non invasive groups; Spearman correlation analysis kn was used. The correlation between OSP classification and Ki-67, p53, Ktrans value, Ve value, Kep value histogram parameters and the knosp classification of pituitary adenoma, the correlation with Ki-67, p53, using the receiver operating characteristics (receiver operating characteristic) curve to determine the maximum diameter of the tumor. Breaking ability; using the Logistic regression analysis model, and using the ROC curve to determine the difference in the diagnostic ability of the two, the difference was statistically significant. Results: 1. the maximum, average, fiftieth percentile, seventy-fifth percentile, ninetieth percentile, ninth of the invasive and non invasive pituitary adenoma's histogram parameters. 500 quantiles, range, mean difference, bias, consistency, energy value, entropy value were significantly different among the two groups (P=0.016,0.002,0.025,0.010,0.006,0.003,0.012,0.008,0.025,0.012,0.028,0.024); the effectiveness of using average, ninety-fifth percentile, and ninetieth percentile to identify invasive and noninvasive pituitary adenomas was better, AUC was 0.792,0.784,0.766; 2. the maximum, average, ninety-fifth percentile, range, standard deviation, variance, relative difference, deviation, kurtosis, conformance, energy value and entropy value were significant differences between the two groups (P= 0.006,0.036,0.023,0.013,0.029,0.027,0.005,0.042,0.013,0.022,0.033,0.046) in invasive and noninvasive pituitary adenomas. The effectiveness of the two persons with relative difference, maximum value and range identification was better, AUC was 0.772,0.766,0.741, and ninetieth percentile, ninety-fifth percentile, relative difference and bias between the two groups of invasive and non invasive pituitary adenomas were significantly different between the two groups (P =0.031,0.013,0.042,0.005,0.023); relative deviation, bias, ninety-fifth were used. The effectiveness of the percentile identification of the two was better, AUC was 0.769,0.740,0.737, the average number of 4.Ktrans, the relative difference of Ve, the combined variable of Kep, the combined variable of the combined variable 1 and the maximum diameter of the tumor were 2, and the AUC of the invasive and non invasive pituitary adenomas was identified by the ROC curve, and the 5. pituitary gland respectively. There was a good correlation between the knosp classification of adenoma and Ki-67, p53 (r=0.547, P0.001; r=0.617, P0.001). 6. the correlation between the mean value of the Ktrans value histogram in the DCE-MRI parameter histogram of the pituitary adenoma and the Knosp classification was the best (r=0.660, p0.001). The best correlation with p53 (r=0.388, p=0.018). Conclusion: the histogram of 1.DCE-MRI parameters can reflect the microvascular heterogeneity of invasive pituitary adenomas and is helpful for the identification of invasive pituitary adenomas and non-invasive pituitary adenomas; 2. the combined use of the DCE-MRI parameter histogram and the maximum diameter of the tumor can improve the diagnosis of invasive pituitary adenomas; 3. pituitary glands. Tumor Kep histogram kurtosis and Ktrans histogram average difference were correlated with Ki-67 and p53 respectively. They could be used as imaging indicators for monitoring the evolution of the disease.

【学位授予单位】：大连医科大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R736.4;R445.2

【参考文献】