卷积神经网络诊断阿尔兹海默症的方法

发布时间：2018-07-18 08:14

【摘要】：针对阿尔兹海默症(AD)通常会导致海马体区域萎缩的现象,提出一种使用卷积神经网络(CNN)对脑部磁共振成像(MRI)的海马体区域进行AD识别的方法。测试数据来自ADNI数据库提供的188位患者和229位正常人的脑部MRI图像。首先,将所有脑图像进行颅骨剥离,并配准到标准模板;其次,使用线性回归进行脑部萎缩的年龄矫正;然后,经过预处理后,从每个对象的3D脑图像的海马体区域提取出多幅2.5D的图像;最后,使用CNN对这些图像进行训练和识别,将同一个对象的图像识别结果用于对该对象的联合诊断。通过多次十折交叉验证方式进行实验,实验结果表明所提方法的平均识别准确率达到88.02%。与堆叠自动编码器(SAE)方法进行比较,比较结果表明,所提方法在仅使用MRI进行诊断的情况下效果比SAE方法有较大提高。
[Abstract]:According to the phenomenon that Alzheimer's disease (AD) usually results in atrophy of hippocampal area, a novel method for AD recognition of hippocampal region in brain magnetic resonance imaging (MRI) using convolution neural network (CNN) is proposed. The data were obtained from MRI images from 188 patients and 229 normal subjects from the ADNI database. First, all brain images were removed from the skull and matched to the standard template. Secondly, linear regression was used to correct the age of brain atrophy. A number of 2.5D images were extracted from the hippocampal region of 3D brain images of each object. Finally, CNN was used to train and recognize these images, and the recognition results of the same object were used for the joint diagnosis of the same object. The experimental results show that the average recognition accuracy of the proposed method is 88.02. Compared with the stacked automatic encoder (SAE) method, the results show that the proposed method is more effective than the SAE method in diagnosis only with MRI.
【作者单位】：福州大学物理与信息工程学院;福建省医疗器械与医药技术重点实验室;厦门理工学院光电与通信工程学院;福建省生态产业绿色技术重点实验室;
【基金】：福建省自然科学基金资助项目(2016J05157) 福建省中青年教师教育科研项目(JAT160074) 厦门市科技计划项目(3502Z20153017)~~
【分类号】：R749.16;TP183

【相似文献】