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: | Shane O'Connell |
| Institution: | NUI Galway, Ireland |
| Department: | Mathematics |
| Country: | |
| Proposed Analysis: | Genome wide association studies of complex polygenic neuropsychiatric disorders often require large sample sizes to be sufficiently powered, owing in part to substantial phenotypic heterogeneity. When significantly associated single nucleotide polymorphisms are identified, interpreting their role in disease aetiology can be difficult considering an overly-broad patient categorisation of 'case' or 'control'. Therefore, we seek to investigate the genetics of Alzheimer's disease using intermediate phenotypes derived from structural neuroimaging data. We hypothesize that this approach can increase the interpretability of the resultant single nucleotide polymorphisms through their association with a replicable and robust neuroanatomical pattern. To extract these intermediate phenotypes, we propose the use of a convolutional neural network, a deep learning architecture specifically designed for image processing applications. Briefly, we will train a convolutional neural network (CNN) classifier that can demonstrate strong discriminative performance, and use saliency-based methods to determine specific regions that are the most influential for model performance. These "important" regions can be thought of as areas relevant to the model's internal representation of the classification problem, and can serve as intermediate phenotypes in a genome wide association study, where we hypothesize that the problem space refinement will facilitate greater interpretation and help to detect previously unreported genetic associations. We have previously applied this CNN approach in bipolar disorder where it outperforms current atlas-based region-of-interest approaches and are now seeking to test its applicability to Alzheimer's disease data. |
| Additional Investigators |
