精神分裂症的MRI图像分类方法研究

A study on MRI image classification of schizophrenia

为实现精神分裂症脑疾病的计算机辅助诊断治疗,将机器学习技术和结构磁共振成像(structural magnetic resonance imaging,sMRI)分析方法应用其中。首先,对sMRI图像库中精神分类症患者和正常人两类受试者进行统计学分析,随后对三维形式的sMRI灰质图像进行切片化及加权平均处理;其次,利用机器学习方法对处理后的图像提取灰度共生矩阵纹理特征;最后,使用XGBoost等算法进行分类,获得精神分裂症诊断结果。实验结果表明,XGBoost算法取得了优秀的诊断精度,为精神分裂症患者的临床诊断与治疗提供生物学依据。

The machine learning techniques and structural Magnetic Resonance Imaging(sMRI)methods are used in computer-aided diagnosis for schizophrenia.First,the two objects of the sMRI image library:schizophrenia patients and normal people,are analyzed statistically.In addition,the slicing and weighted averaging method is proposed for gray image of the sMRI in the form of three-dimensional volume data.The features about gray matter images of sMRI are extracted and classified.Secondly,texture feature of gray-level co-occurrence matrices is extracted from the above processed images.Finally,the XGBoost classifier is used for classification.Experiments show that the diagnostic accuracy of the proposed method is good,and a biological basis for the clinical diagnosis and treatment of schizophrenia is provided.