基于活动邻域模型的脑组织提取及参数自动设定

Active Neighborhood Model Based Brain Extraction Method and Automatic Parameter Setting

从T1加权MRI图像中提取脑组织是脑容量测定、脑组织分割、脑萎缩诊断等的重要预处理过程。本文将随机森林法引入基于活动邻域模型的脑组织提取方法中以实现参数的自动设置。该方法先对训练样本进行脑组织提取和特征提取,根据相应参数采用随机森林法对样本进行分类,对于测试样本进行类别判别,根据该类别实现自动设定参数。通过验证证实,本研究所设计的脑组织提取方法具有较好的鲁棒性,设计的参数自动设定方法是可行的,能减少用户交互。

The extraction of brain tissue from cerebral T1-Weighted MRI volume is an important pre-procedure for determination of brain volume,segmentation of brain tissues,and quantification of brain atrophy. The random forest method was used in the active neighborhood model based brain extraction method to set the parameters automatically. This method firstly performed brain extraction on training samples and obtained the corresponding parameters. After feature extraction of training samples,this paper used random forest classifier to classify the training samples and then performed discriminant for testing samples to obtain the parameters of testing samples. Experiments show our brain extraction method has good robustness,and the proposed parameter setting method is feasible and can reduce the interaction of the users.