Deep 3D Convolutional Neural Network Architecture for Alzheimer's Disease Diagnosis

• Main Authors: Hiroki Karasawa, 唐澤大樹
• Other Authors: Chien-Liang Liu
• Format: Others
• Language: en_US
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/7v9d9v

碩士 === 國立交通大學 === 工業工程與管理系所 === 106 === Recently, dementia has become a social problem in advanced country as an aging society. Currently, there are 46.8 million people with dementia worldwide, and it is predicted to be 130 million people as threefold in 2050. Alzheimer’s disease (AD) is the most common case of dementia. The cost of care for AD patients in 2015 is about 818 billion US dollars, and the cost is expected to increase dramatically in the future due to the increase in the number of patients due to the aging society. However, it is still very difficult to cure AD, explaining why the detection of AD plays an important role. This thesis proposes to use machine learning to detect AD from brain image data. Once the detection of AD is possible, it will be possible to apply different medical treatments to patients to prevent from AD. Most machine learning algorithms rely on good feature representations, which are commonly obtained manually and require domain experts to provide guidance. Apparently, feature extraction is a time-consuming and labor-intensive task. In contrast, 3D Convolutional Neural Network (3DCNN) automatically learns feature representation from images, and is not much affected by image processing. However, the performance of CNN highly depends on its layer architecture. This thesis proposes a novel 3DCNN architecture for MRI image diagnosis of AD, aiming to give accurate diagnosis for AD patients. The experimental results indicate that the proposed model performs very well on the detection of AD.