基于多模态磁共振影像的首发未用药精神分裂症自动分类研究

发布时间：2018-08-25 10:33

【摘要】：慢性精神分裂症患者大脑的结构和功能异常已经被广泛报道,但是首发未用药精神分裂症患者和正常人的相关研究较少。本研究采集了44名首发未用药精神分裂症患者和56名正常人的结构和静息态功能磁共振图像,基于自动解剖标签模板提取了90个感兴趣区域的灰质体积、局部一致性、低频振荡振幅和度中心度作为特征,并将这些特征作为输入,用基于递归特征消除的支持向量机对首发未用药精神分裂症患者和正常人进行分类。结果表明,局部一致性和低频振荡振幅的组合为最佳分类特征,分类准确率达到96.97%,并且分类权重最大的脑区主要位于额叶。研究结果有利于加深对精神分裂症神经病理机制的了解,有助于开发出用于临床辅助诊断的生物学标记物。
[Abstract]:The abnormal structure and function of the brain in patients with chronic schizophrenia have been widely reported, but few studies have been carried out in the first batch of patients with untreated schizophrenia and normal people. In this study, the structural and resting functional magnetic resonance images of 44 first-episode schizophrenic patients and 56 normal subjects were collected. The gray matter volume and local consistency of 90 regions of interest were extracted based on automatic anatomical label template. Low frequency oscillation amplitude and degree of centrality are used as features and these characteristics are used as input to classify first-episode schizophrenic patients and normal people by support vector machine based on recursive feature elimination. The results show that the combination of local consistency and low frequency oscillation amplitude is the best classification feature and the classification accuracy is 96.97 and the most weighted brain area is mainly located in the frontal lobe. The results are helpful to further understand the neuropathic mechanism of schizophrenia and to develop biological markers for clinical diagnosis.
【作者单位】：华南理工大学医学院;华南理工大学材料科学与工程学院生物医学工程系;广东省人民医院放射科;广州医科大学附属脑科医院(广州市惠爱医院)/广州市脑科医院精神科;广州医科大学附属脑科医院/广州市脑科医院-华工精神影像联合研究中心;
【基金】：国家自然科学基金青年科学基金项目(31400845);国家自然科学基金面上项目(81471654,81571333,31771074) 广东省自然科学基金资助项目(2015A030313800) 广东省前沿与关键技术创新专项资金(重大科技专项)(2016B010108003) 广东省公益研究与能力建设专项资金(2016A020216004) 广东省协同创新与平台环境建设专项资金(2017A040405059) 广州市产学研协同创新重大专项(201604020170,201704020168,201704020113) 华南理工大学中央高校基本科研业务费(2015ZZ042) 广州市医学重点学科建设广州市惠爱医院课题(GBH2014-QN06)
【分类号】：R445.2;R749.3;TP391.41

【相似文献】