Predicting Conversion of Mild Cognitive Impairment to Alzheimer's Disease by Modelling Healthy Ageing Trajectories

Alzheimer's disease (AD) is a heterogeneous type of dementia. Mild cognitive impairment (MCI) is the early stage of cognitive decline before AD. Predicting the MCI-to-AD conversion remains challenging due to the heterogeneity of patients. Previous evidence shows that the brain network generated from diffusion MRI promises to classify dementia. However, the limited availability of diffusion MRI obstructs the training of deep learning models. Here, we firstly develop a cross-modal approach to generate brain networks from anatomical MRI under the guidance of diffusion MRI. We then construct the healthy ageing trajectories of controls with a generative model. Finally, we design a recurrent neural networks-based approach to model the longitudinal deviation of patients' brain networks from the healthy ageing trajectory. Numerical results show that the proposed methods outperform the benchmarks in the prediction task. We also visualize the model interpretation to explain the prediction and identify abnormal changes in white matter tracts.