MRI Based Spatio-Temporal Model for Alzheimer’s Disease Prediction

Alzheimer’s disease (AD) is a leading cause of memory loss and eventually leads to the mortality of affected individuals. It is characterized by structural changes in the brain, including the shrinkage of specific regions and the development of abnormalities. AD is a progressive neurodegenerative disease that worsens over time. Spatial analysis helps identify the affected brain regions, while temporal analysis provides insights into the disease progression and temporal patterns of change. Integrating both spatial and temporal information allows for a more accurate and detailed characterization of AD. Consequently, the development of a spatio-temporal model holds promise for early diagnosis of AD. This work proposes a spatio-temporal analysis using deep neural networks this utilizes Magnetic Resonance Imaging (MRI) data obtained from two sources: the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database and the Open Access Series of Imaging Scans (OASIS). Demographic information such as Mini-Mental State Examination (MMSE), Clinical Dementia Rating (CDR), Global Clinical Dementia Rating (GCDR), and whole brain volume are important in Alzheimer’s prediction, as they provide valuable insights into cognitive function, disease progression, and brain health, aiding in the identification and assessment of individuals at risk or in the early stages of Alzheimer’s disease. A spatio-temporal model called Convolutional Long Short-Term Memory (ConvLSTM) was developed for the analysis of spatio-temporal features using deep neural networks. The proposed ConvLSTM-based spatio-temporal model achieves an accuracy of 98% for all types of datasets, both with and without demographic information, enabling early prediction of AD.