中文摘要：

阿尔茨海默病（AD）症不同状态所对应的脑部年龄与实际年龄存在偏差,为了准确预测该偏差得到脑部年龄,在现有基于脑磁共振（MR）图像年龄检测方法的基础上,提出了一种新的脑部年龄检测算法.首先,设定偏差搜索范围;然后,基于可分性距离判据设计了适应度函数,利用偏差和支持向量回归机（SVR）获得样本年龄估计值,并计算出其适应度值;接着,通过最大化适应度值获取偏差的最优估计,从而获得更有利于AD症分类的脑部年龄.与现有的年龄检测方法相比,对于正常对照组（NC）与AD患者、NC与轻度认知障碍患者（MCI）以及MCI与AD患者3种分类情况,所提方法的可分度值分别提高了0.178,0.033,0.017.因此,所提方法检测的年龄具有更好的可分性,更有利于提高AD症的分类准确率.

英文摘要：

There is a deviation between brain age and actual age corresponding to different states of Alzheimer’s disease（AD）. In order to accurately predict the deviation to obtain brain age, a new method of brain age detection is proposed based on the existing magnetic resonance （MR） image age detection methods. First, the deviation search range is set. Secondly, a fitness function is designed based on the distance separability criterion, and the brain age of samples is estimated via deviation and support vector regression（SVR） and the fitness value is calculated. Thirdly, the optimal deviation is obtained by maximizing the fitness value, so that obtaining brain ages more conducive to the classification of AD. Finally, the proposed method is compared with the existing age detection method. For three kinds of classification which are normal control group （NC） and AD, NC and mild cognitive impairment （MCI） as well as MCI and AD, based on the proposed algorithm, the separability can be improved by 0.178, 0. 033, and 0.017, respectively. Therefore, the age detected with the proposed algorithm has better separability and helps to improve the classification accuracy of AD.