论文部分内容阅读
注意缺陷多动障碍(ADHD)是一种高发于学龄儿童的行为障碍综合症。目前,ADHD的诊断主要依赖主观方法,导致漏诊率和误诊率较高。基于此,本文提出一种基于卷积神经网络的ADHD客观分类算法。首先,对脑部磁共振图像(MRI)进行头骨剥离、高斯核平滑等预处理;其次,对大脑的右侧尾状核、左侧楔前叶和左侧额上回部位的MRI进行粗分割;最后,利用3层卷积神经网络进行分类。实验结果表明:1本文的算法能有效地对ADHD和正常人群进行分类;2右侧尾状核和左侧楔前叶的ADHD分类准确率要高于ADHD-200全球竞赛中所有方法达到的ADHD最高分类准确率(62.52%);3利用上述3个脑区对ADHD患者和正常人群进行分类,其中右侧尾状核的分类准确率最高。综上所述,本文提出了一种利用粗分割和深度学习对ADHD患者和正常人群进行分类的方法。本文方法分类准确率高,计算量小,能较好地提取不明显的图像特征,改善了传统MRI脑区精确分割耗时长及复杂度高的缺点,为ADHD的诊断提供了一种可参照的客观方法。
Attention Deficit Hyperactivity Disorder (ADHD) is a disorder of behavior disorder that occurs in school-age children. At present, the diagnosis of ADHD mainly depends on subjective methods, resulting in a higher rate of missed diagnosis and misdiagnosis. Based on this, this paper presents an ADHD objective classification algorithm based on convolutional neural network. Firstly, skull dissection, Gaussian nucleus smoothing and other preprocessing were performed on magnetic resonance imaging (MRI) of the brain. Secondly, MRI was performed on the right caudate nucleus, the left anterior wedge and the left superior frontal gyrus Finally, we use the 3-layer convolution neural network to classify. Experimental results show that: 1 The algorithm in this paper can effectively classify ADHD and normal population; 2 ADHD classification accuracy of the right caudate and left anterior wedge leaves is higher than ADHD achieved by all methods in the ADHD-200 global competition The highest classification accuracy (62.52%); 3 using the above three brain regions ADHD patients and normal population classification, including the right side of the highest classification accuracy of caudate. In summary, this paper presents a method of using coarse segmentation and depth learning to classify patients with ADHD and normal population. The proposed method has the advantages of high accuracy and small amount of calculation, which can extract obvious image features better and improve the shortcomings of traditional MRI brain segmentation with high precision and time-consuming. It provides a reference for the diagnosis of ADHD Objective method.