关键词: 关键词: 阿尔兹海默症, 卷积神经网络, 磁核共振图像, 海马体, 计算机辅助诊断

Abstract: Abstract: Alzheimer's disease usually leads to atrophy of the hippocampus region, a Convolutional Neural Network(CNN) based method was used for diagnosis of Alzheimer's Disease(AD) using hippocampus region of brain Magnetic Resonance Imaging(MRI). All data were from the ADNI database including 188 AD and 229 Normal Control(NC). MR images were first preprocessed by skull stripped and aligned to a template space. Then a linear regression model was used for age correction, and multi 2.5D pictures were extracted from hippocampus region with MRI. Finally, a CNN was used for diagnosis of AD with input of these 2.5D pictures. Through repeated 30 times ten fold cross validation, the mean accuracy was 88.02%. The proposed method has a acceptable diagnosis result when using only MRI for diagnosis Alzheimer's disease.

Key words: Keywords: Alzheimer's Disease, Convolutional Neural Network, Magnetic Resonance Imaging, Hippocampus, Computer-aid diagnosis