基于结构核磁共振影像，探讨轻度认知障碍向阿尔兹海默病转化情况.选取152例轻度认知障碍、101例阿尔兹海默病、97例正常对照组的结构核磁共振影像为研究对象.首先，采取独立成分分析方法对阿尔兹海默病和正常对照样本进行组间特征提取；其次，执行感兴趣体素分析，获取组间差异特征信息，并将轻度认知障碍样本在阿尔兹海默病和正常对照组的特征空间上进行表达，获取轻度认知障碍样本特征信息；最后，运支持向量机，对轻度认知障碍样本进行转化预测.测试结果：预测准确率、灵敏度、特异度分别达到80.26%、86.30%、74.68%.结果显示，利用阿尔兹海默病和正常对照个体之间的差异特征能够比较准确地对轻度认知障碍向阿尔兹海默病转化进行预测.%The paper discusses the conversion from mild cognitive impairment to Alzheimer's disease based on structural magnetic resonance imaging. 152 mild cognitive impairment patients, 101 Alzheimer's disease patients and 97 normal controls are selected as study objects. Firstly, independent component analysis is adopted to execute feature extraction of Alzheimer's disease and normal controls samples. Then, selection of voxel of interest is executed to get different feature information between two groups. And feature information of mild cognitive impairment is obtained by putting mild cognitive impairment samples to the feature space of Alzheimer's disease and normal controls samples. Finally, support vector machine is employed to predict whether mild cognitive impairment will convert to Alzheimer's disease. Prediction accuracy, sensitivity and specificity are 80. 26%, 86. 30%, and 74. 68% respectively. The results show that discriminative features between Alzheimer's disease patients and normal controls can accurately predict whether mild cognitive impairment will convert to Alzheimer's disease.
展开▼
