基于原型学习与深度特征融合的脑功能连接分类方法研究

Brain Functional Connection Classification Method Based on Prototype Learning and Deep Feature Fusion

近年来,基于深度学习的脑功能连接分类方法已成为一个研究热点.为了进一步提高脑功能连接的分类准确率,获得与疾病相关的鉴别性特征,本文提出了一种基于原型学习与深度特征融合的脑功能连接分类方法.该方法首先使用栈式自编码器从脑功能连接中提取从低层次到高层次的深度特征;然后利用原型学习在自编码器的各隐层中提取表示样本类别信息的距离特征;最后采用深度特征融合策略将这些距离特征融合,并将该融合特征用于脑功能连接的类别标签预测.在ABIDE数据集上的实验结果表明,与其他同类方法相比,该方法不仅具有较高的分类准确率,而且能够更加准确地定位与疾病相关的脑区.

In recent years,the brain functional connection classification method based on deep learning has become a hot research topic.In order to further improve the classification accuracy of brain functional connections and gain discriminative features associated with a disease,we propose a brain functional connection classification method based on prototype learning and deep feature fusion in this paper.Firstly,we use stacked autoencoders to extract lower-to-higher deep features from brain functional connections.Then the prototype learning is used to extract the distance features of the sample category from each hidden layer of the stacked autoencoders.Finally,the deep feature fusion strategy is adopted to fuse these distance features and the fused features are applied for the brain functional connection classification.The experimental results on the ABIDE dataset show that compared with other methods,the proposed method not only has a higher classification accuracy,but also can locate the brain areas related to the disease more precisely.