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: | Qihang Yao |
| Institution: | Georgia Institute of Technology |
| Department: | Computer Science |
| Country: | |
| Proposed Analysis: | In general, we want to show the effectiveness of TRACED in locating structural differences that are important to brain disease. We want to do so by showing that the structural differences located by TRACED are a set of good features to build a classification model to discriminate between diseased and control subjects. ### What is TRACED? * Infer activation cascades (that resemble functional activity) from structural connectomes. * Find significant activation cascade differences between groups of subjects with some qualitative differences (e.g. MDD versus control) * Locate the key structural differences that lead to functional differences. ### How do we validate TRACED with Machine Learning? * If TRACED identifies some structural difference in connectomes between two groups of subjects, can we argue that these differences are indeed helpful for discriminating between two groups? * Can we show that these structural differences, when used as features, can lead to close-to-optimal classification performance given the connectome? * In other words, we assume that there is an upper bound of classification performance on a certain dataset of connectomes containing two groups of subjects. * This upper bound is because of the number of samples that we can have, heterogeneity and noises in the data. * Can we get close to this upper bound much easier with TRACED as the guide? |
| Additional Investigators |
