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: | 晏丞 卓 |
| Institution: | Department of Mathematics National Chung Cheng University |
| Department: | Statistical science |
| Country: | |
| Proposed Analysis: | Neuroimaging data are often collected as a spatial array of numbers with each array element corresponding to an observed data value at a particular brain location. Model selection becomes an active research topic but has’t been explored largely due to the complex spatial correlation structure of the neuroimaging data set. To the best of our knowledge, a formal statistical procedure for mean model and spatial correlation structure determination has been underdeveloped in literature, especially for a distribution-free regression framework. To address this important issue, we develop a joint model selection criterion JSIC (Joint Spatial Information Criterion) for simultaneously selecting the marginal mean regression and the correlation/covariance structure for spatially correlated data when there are large numbers of spatially correlated observations on a moderate number of subjects. Also, our simulation results show that the proposed model selection criterion applies well in both continuous and binary response data. In the example of Albert and McShane (1995) , CT (computer tomography) scans were collected on a large group of stroke patients for the purpose of examining the characteristics of stroke-induced lesions. The goal of our study is to build up a predictive regression model for the lesion frequency with spatial location and subject-specific covariates. |
| Additional Investigators | |
| Investigator's Name: | 孟修 吳 |
| Proposed Analysis: | Neuroimaging data are often collected as a spatial array of numbers with each array element corresponding to an observed data value at a particular brain location. Model selection becomes an active research topic but has’t been explored largely due to the complex spatial correlation structure of the neuroimaging data set. To the best of our knowledge, a formal statistical procedure for mean model and spatial correlation structure determination has been underdeveloped in literature, especially for a distribution-free regression framework. To address this important issue, we develop a joint model selection criterion JSIC (Joint Spatial Information Criterion) for simultaneously selecting the marginal mean regression and the correlation/covariance structure for spatially correlated data when there are large numbers of spatially correlated observations on a moderate number of subjects. Also, our simulation results show that the proposed model selection criterion applies well in both continuous and binary response data. In the example of Albert and McShane (1995) , CT (computer tomography) scans were collected on a large group of stroke patients for the purpose of examining the characteristics of stroke-induced lesions. The goal of our study is to build up a predictive regression model for the lesion frequency with spatial location and subject-specific covariates. |
