摘要
动态脑网络能有效反映脑网络中连接结构的动态变化信息,被广泛使用于脑疾病的识别研究中。动态脑网络由一组连接矩阵组成。通常研究者会基于矩阵上三角元素向量的L2距离,计算所有样本连接矩阵的距离矩阵,使用状态聚类将这些连接矩阵划分为不同的状态。但是简单地使用L2距离,且在全部样本上进行状态聚类会导致忽视连接矩阵所代表的脑网络的图结构信息以及个体之间的差异。因此,提出一种新的基于图核的动态脑网络状态构建方法。该方法针对单个体的动态脑网络设计,使用图核衡量单个样本的动态脑网络连接矩阵之间的相似性,随后根据相似性矩阵,将连接矩阵与其最相似的矩阵进行合并。在精神分裂症数据集上验证该方法的有效性,其结果证明所提方法可以获取81.6%的分类精度。
The dynamic brain network can effectively reflect the dynamic information of the structure in the brain network,and is widely used in the research of brain disease identification.The dynamic brain network consists of a set of connection matrix.Generally,researchers calculated the distance matrix of all samples'connection matrices based on the L2 distance of the vector generated by the triangular elements of the matrix,and then used K-means clustering to divide these connection matrices into different states.However,simple using the L2 distance and K-means clustering on all samples would lead to ignore the graph structure information of brain network represented by the connection matrix and individual differences between samples.Hence,to solve these two problems,this paper proposed a new dynamic brain network state construction method based on the graph kernel.Aiming at the construction of the dynamic brain network states for each single sample,this method used the graph kernel to measure the similarity between connection matrixes of the dynamic brain network,and combined the connection matrix and it's the most similar matrix according to the similarity matrix.We extensively evaluated this method on a real Schizophrenia dataset.The results show that the proposed method can obtain 81.6%classification accuracy.
作者
袁新颜
黄嘉爽
Yuan Xinyan;Huang Jiashuang(Jiangsu Vocational College of Business,Nantong 226011,Jiangsu,China;School of Information Science and Technology,Nantong University,Nantong 226019,Jiangsu,China)
出处
《计算机应用与软件》
北大核心
2023年第12期108-113,168,共7页
Computer Applications and Software
基金
南通市科技局基础科学研究项目(JC22022060)。
关键词
动态脑网络
精神分裂症
图核
网络状态
Dynamic brain network
Schizophrenia
Graph kernel
Network state
