基于极限学习机的阿尔兹海默病辅助诊断

阿尔兹海默病是一种渐进发展式的痴呆疾病, 其脑部随着病情发展逐渐出现萎缩。利用磁共振脑图像解剖学特征的变化, 提出一种使用极限学习机来诊断阿尔兹海默病以及轻度认知障碍的方法。采用FreeSurfer软件, 分析从ADNI数据库的818份磁共振图像中得到的脑部解剖学特征。首先对这些特征使用线性回归模型来估计正常衰老引起的萎缩因素, 并将其从特征中去除;随后采用极限学习机作为分类器, 使用处理后的特征来诊断阿尔兹海默病和轻度认知障碍。在实验过程中, 通过十折交叉验证来测试该方法的诊断准确率、敏感度、特异度和曲线下面积。通过100次实验求平均的方式计算得出, 该方法诊断阿尔兹海默病的准确率达到87.62%, 曲线下面积达到94.25%;诊断轻度认知障碍的准确率达到73.38%, 敏感度达到83.88%, 其中年龄矫正能有效提高轻度认知障碍诊断的准确率。实验结果表明, 该方法能有效诊断阿尔兹海默病和轻度认知障碍。

Computer-Aided Diagnosis of Alzheimer's Disease Based on Extreme Learning Machine

Alzheimer's disease is a progressive disease of dementia usually associated with brain atrophy. We proposed a method of diagnosis of Alzheimer's disease (AD) and mild cognitive impairment (MCI) with the anatomical features of MRI brain images. Method: The data were obtained from ADNI dataset, and the anatomical features of 818 subjects were computed by FreeSurfer software, these features were first preprocessed with age correction algorithm using linear regression to estimate normal aging effect, and was then removed from features. The extreme learning machine was utilized as classifier for diagnosis of AD and MCI with these preprocessed features. The ten-fold cross validation was adopted for calculating accuracy, sensitivity, specificity and area under curve (AUC). Results: By making average with 100 runs, the accuracy of diagnosis of AD was 87.62%, and the AUC reached 94.25%. The accuracy of diagnosis of MCI was 73.38%, and the sensitivity reached 83.88%. The age correction can improve the accuracy of MCI diagnosis. The results demonstrated the efficacy of the proposed method for diagnosis of AD and MCI.