There are many active research projects accessing and applying shared ADNI data. Use the search above to find specific research focuses on the active ADNI investigations. This information is requested annually as a requirement for data access.
| Principal Investigator | |
| Principal Investigator's Name: | Kevin Leung |
| Institution: | Johns Hopkins University |
| Department: | Biomedical Engineering |
| Country: | |
| Proposed Analysis: | Alzheimer’s disease (AD) is one of the most common forms of dementia and an irreversible progressive brain disorder that is characterized by a decline in cognitive function and memory deficits. There has been a growing interest in machine learning-based methods for the analysis of neurodegenerative disorders using multimodal neuroimaging data. However, the pre-processing procedures, including feature extraction and selection, of traditional machine learning methods may require specialized knowledge and expertise, may be time-consuming, and may have issues with reproducibility. To address this challenge, deep learning methods, such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs), have been developed to perform automatic feature extraction and have shown promise in disease classification and prognostic prediction tasks for neurodegenerative disorders. However, deep learning methods may also suffer from high variance in prediction due to the nonlinearity of neural networks and high model complexity. Ensemble learning methods have been used to improve the accuracy of prediction tasks by combining multiple classifiers to reduce the variance in prediction. We have previously developed a three-stage, deep learning, ensemble approach for prognosis in Parkinson’s disease (PD) and validated this approach using longitudinal clinical data from the Parkinson’s Progression Marker’s Initiative (PPMI). The ensemble approach was developed to predict longitudinal motor outcome in PD 4 years after initial baseline screening by incorporating imaging and non-imaging measures from baseline and 1 year after baseline. We aim to leverage our approach to achieve improved prognosis in patients with AD. Specifically, we aim to develop an ensemble deep learning framework for prognosis of patients with AD using imaging features from longitudinal multimodal PET/MRI scans as well as non-imaging clinical measures as inputs to perform prognostic prediction in AD. |
| Additional Investigators |
