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: | Patrick Henriksen |
| Institution: | Imperial College London |
| Department: | Department of Computing |
| Country: | |
| Proposed Analysis: | The overarching goal of this project is to analyse whether the performance of neural networks trained on MRI data may be improved by applying formal robust training methods. Specifically, we will train neural networks using methods for certifiable robust training such as CROWN-IBP (ICLR 2020) for white-noise perturbations and SSIP-IBP (AAAI 2023) for bias field perturbations. We will then explore whether the resulting networks: 1) exhibit improved performance compared to networks trained with standard training, especially in semi-supervised learning and small-data settings; 2) exhibit improved performance compared to networks trained with non-formal robustness-promoting methods such as adversarial augmentation; 3) generalise better to slight variations in the input data induced by, e.g. different scanners. |
| Additional Investigators |
