基于GMWMVGG13-SVM的精神分裂症辅助诊断方法

ASSISTED DIAGNOSIS OF SCHIZOPHRENIA BASED ON GMWMVGG13-SVM

精神分裂症是一种多发于青年人的脑疾病,尽早诊断治疗能减轻家庭负担以及减少社会成本。对此,提出一种精神分裂症计算机辅助诊断方法。对精神分裂症sMRI数据集进行形态学分析、无信息切片筛除和特征提取器输入拟合;迁移VGG13提取白质(White matter,WM)特征、灰质(Gray matter,GM)特征,将Inception-ResNet V2作为对比模型,提取相同两种特征;经PCA降维的GM、WM和GMWM特征在网格搜索算法作用下训练最佳SVM分类器。实验结果表明,在诊断精神分裂症sMRI应用下,该方法比现有方法的准确率提高了约6百分点。

Schizophrenia is a brain disease that occurs in young people.Early diagnosis and treatment can reduce the burden on families and reduce social costs.This paper proposes a novel computer-assisted diagnosis method for schizophrenia.The sMRI data set was preprocessed,mainly for morphological analysis,no information slice filtering and feature extractor format fitting;the Gray matter(GM)features and White matter(WM)features were extracted by VGG13,and inception-resnet V2 was used as a comparison model for extracting features,which extracted the same two features;the GM,WM and combined features of PCA dimensionality reduction were used to train the best SVM classifier under the action of grid search algorithm.The experimental results show that our method improved the accuracy of sMRI diagnosis of schizophrenia by about 6 percentage points.