Alzheimer's Disease Diagnosis Model Based on Three-Dimensional Full Convolutional DenseNet

• Published in: 2019 10th International Conference on Information Technology in Medicine and Education (ITME) pp. 13 - 17
• Main Authors: He, Guangyu, Ping, An, Wang, Xi, Zhu, Yufei
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.08.2019
• Subjects: Alzheimer's disease; artificial intelligence; Convolutional neural networks; Data models; deep learning; Magnetic resonance imaging; medical diagnosis; medical image analysis; Solid modeling; Training
• ISSN: 2474-3828
• DOI: 10.1109/ITME.2019.00014

Dementia has become one of the major diseases causing death and disability in the elderly. Alzheimer's disease is the primary cause of dementia, and its clinical diagnosis, especially early diagnosis, is very urgent. The research on the diagnosis of Alzheimer's disease based on deep convolutional neural network has made good progress. However, it is still difficult to meet the requirements of clinical application. In clinic, most of them are structural MRI images, with limited sample size and few scanning layers. But deep learning needs enough annotated data. Aiming at the practical needs of clinical diagnosis of Alzheimer's disease, this paper proposes a increment method of dataset by weighted combination of positive and negative samples and a learning method for small number of samples, and establishes a 3D full convolutional DenseNet classification model, which can not only obtain better image feature information, but also improve the generalization ability of the model.