常规MRI图像纹理分析对颞叶癫痫海马硬化的诊断价值

摘要

目的 探讨常规MRI图像纹理分析对颞叶癫痫海马硬化的诊断价值.材料与方法 应用Mazda软件分析22例病理证实为颞叶癫痫海马硬化患者的术前斜冠状位海马T2 FLAIR图像纹理.采用3种特征选择方法[Fisher系数(fisher coefficient,Fisher)、分类错误概率联合平均相关系数(classification error probability combined with average correlation coefficients,POE+ACC)、交互信息(mutual information,MI),3种方法联合简称为FPM]提取患者双侧海马的纹理特征;4种统计方法[线性判别分析(linear discriminant analysis,LDA)、非线性判别分析(nonlinear discriminant analysis,NDA)、原始数据分析(raw data analysis,RDA)、主要成分分析(principal component analysis,PCA)]判别患者海马硬化侧与正常侧,结果 以错判率表示.同时由2名影像医师对22例患者的MR图像进行诊断.比较两种判断结果的差异.结果采用FPM选择方法及NDA统计学方法误判率最小(2/44,4.55%),较影像医师的错判率(11/44,25%)低,差异有统计学意义(P<0.05).结论 常规MRI纹理分析可为颞叶癫痫海马硬化的诊断提供可靠的客观依据.%Objective: To explore the value of conventional MRI texture analysis in diagnosing temporal lobe epilepsy due to hippocampal sclerosis. Materials and Methods: The oblique coronal T2 FLAIR images of 22 patients with temporal lobe epilepsy and hippocampal sclerosis confirmed by pathology were analyzed by using Mazda software. The feature selection methods included mutual information (MI), Fisher coefficient and classification error probability combined with average correlation coefficients (POE+ACC), Through these methods, the texture features of hippocampus were extracted. Then, four statistical methods were used to distinguish the hippocampal sclerosis side and normal side of the patients that were raw data analysis (Raw Date), the principal component analysis (PCA), linear discriminant analysis (LDA) and nonlinear discriminant analysis (NDA). The results are indicated with misclassification rate. Meanwhile, 2 neuroradiologists also reviewed the MR images of 22 patients. The differences of the results between the two analysis methods were analyzed finally. Results: The misclassification rate was the lowest (2/44, 4.55%) via the FPM selection and NDA statistical method. Furthermore, there were statistically significant differences of the misclassification rate between the texture analysis and neuroradiologists' analysis (11/44, 25%). Conclusion: Texture analysis of conventional MRI can provide reliable objective basis for the diagnosis of temporal lobe epilepsy due to hippocampal sclerosis.