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: | Hai Shu |
| Institution: | New York University |
| Department: | Biostatistics |
| Country: | |
| Proposed Analysis: | We propose to develop a spatial FDR controlling method that utilizes the deep convolutional neural network (DCNN), a powerful deep-learning model, to sufficiently capture the spatial dependence among test statistics and thus to minimize the false nondiscovery rate (FNR), a measure of type II error. The DCNN has achieved remark- able success in various computer vision tasks. It is capable of capturing complex spatial dependence by (i) using convolutional filters to incorporate the local correlations among neighboring voxels and (ii) exploiting multiple- layer architecture to learn the global dependence structure like interregional and/or long-range dependencies. We will effectively integrate the DCNN with high-dimensional multiple testing, and expect to obtain the desirable optimality for the proposed FDR method given the existing literature on the local index of significance. We will apply the proposed spatial FDR controlling method to investigate the progression of AD using the FDG-PET imaging data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). We will compare the CMRgl differences between and within the three disease status groups, namely, AD patients, mild cognitive impairment (MCI) patients, and cognitively normal (CN) subjects, across four time points from the baseline up to the 24th month. The proposed FDR method is expected to be more powerful than competing approaches to detect the voxel-level differences in CMRgl, and may potentially discover some unknown brain regions that are vulnerable to AD, thereby facilitating new clinical and pathological research. The proposed spatial FDR controlling method will also be widely applicable to large-scale multiple testing problems in other fields of biomedical research. |
| Additional Investigators | |
| Investigator's Name: | Qiran Jia |
| Proposed Analysis: | Alzheimer’s disease (AD) is the most common type of dementia, accounting for approximately 60–80% of all cases. As a neurodegeneration biomarker, Fluorine-18 fluorodeoxyglucose positron emission tomography (FDG-PET) measures the cerebral metabolic rate of glucose (CMRgl) as a proxy of neural activity, and is extensively used in early diagnosis and monitoring progression of AD. The CMRgl difference between two disease-status groups can be investigated by testing the difference in the mean values of CMRgl at each brain image voxel, thereby becoming a high-dimensional multiple testing problem. The false discovery rate (FDR), a measure of type I error, provides a powerful and practical criterion for large-scale multiple testing problems. However, most FDR controlling approaches used in neuroimaging studies ignore the spatial dependence among the test statistics obtained from brain voxels and thus lose substantial power to effectively and accurately identify AD-related regions. This project aims to conduct a comparative study of existing FDR controlling approaches, in particular, recent spatial FDR methods, on AD's FDG-PET neuroimaging data to discover new and important AD-related brain regions that are missed by conventional FDR methods. We will use the FDG-PET data from the Alzheimer’s Disease Neuroimaging Initiative, and compare the CMRgl differences between the three disease-status groups, including AD patients, mild cognitive impairment patients, and cognitively normal subjects. |
| Investigator's Name: | Taehyo Kim |
| Proposed Analysis: | Alzheimer’s disease (AD) is the most common type of dementia, accounting for approximately 60–80% of all cases. As a neurodegeneration biomarker, Fluorine-18 fluorodeoxyglucose positron emission tomography (FDG-PET) measures the cerebral metabolic rate of glucose (CMRgl) as a proxy of neural activity, and is extensively used in early diagnosis and monitoring progression of AD. The CMRgl difference between two disease-status groups can be investigated by testing the difference in the mean values of CMRgl at each brain image voxel, thereby becoming a high-dimensional multiple testing problem. The false discovery rate (FDR), a measure of type I error, provides a powerful and practical criterion for large-scale multiple testing problems. However, most FDR controlling approaches used in neuroimaging studies ignore the spatial dependence among the test statistics obtained from brain voxels and thus lose substantial power to effectively and accurately identify AD-related regions. This project aims to conduct a comparative study of existing FDR controlling approaches, in particular, recent spatial FDR methods, on AD's FDG-PET neuroimaging data to discover new and important AD-related brain regions that are missed by conventional FDR methods. We will use the FDG-PET data from the Alzheimer’s Disease Neuroimaging Initiative, and compare the CMRgl differences between the three disease-status groups, including AD patients, mild cognitive impairment patients, and cognitively normal subjects. |
| Investigator's Name: | Vanessa Martinez |
| Proposed Analysis: | Alzheimer’s disease (AD) is the most common type of dementia, accounting for approximately 60–80% of all cases. As a neurodegeneration biomarker, Fluorine-18 fluorodeoxyglucose positron emission tomography (FDG-PET) measures the cerebral metabolic rate of glucose (CMRgl) as a proxy of neural activity, and is extensively used in early diagnosis and monitoring progression of AD. The CMRgl difference between two disease-status groups can be investigated by testing the difference in the mean values of CMRgl at each brain image voxel, thereby becoming a high-dimensional multiple testing problem. The false discovery rate (FDR), a measure of type I error, provides a powerful and practical criterion for large-scale multiple testing problems. However, most FDR controlling approaches used in neuroimaging studies ignore the spatial dependence among the test statistics obtained from brain voxels and thus lose substantial power to effectively and accurately identify AD-related regions. This project aims to conduct a comparative study of existing FDR controlling approaches, in particular, recent spatial FDR methods, on AD's FDG-PET neuroimaging data to discover new and important AD-related brain regions that are missed by conventional FDR methods. We will use the FDG-PET data from the Alzheimer’s Disease Neuroimaging Initiative, and compare the CMRgl differences between the three disease-status groups, including AD patients, mild cognitive impairment patients, and cognitively normal subjects. |
| Investigator's Name: | Gabe Grajeda |
| Proposed Analysis: | Alzheimer’s disease (AD) is the most common type of dementia, accounting for approximately 60–80% of all cases. As a neurodegeneration biomarker, Fluorine-18 fluorodeoxyglucose positron emission tomography (FDG-PET) measures the cerebral metabolic rate of glucose (CMRgl) as a proxy of neural activity, and is extensively used in early diagnosis and monitoring progression of AD. The CMRgl difference between two disease-status groups can be investigated by testing the difference in the mean values of CMRgl at each brain image voxel, thereby becoming a high-dimensional multiple testing problem. The false discovery rate (FDR), a measure of type I error, provides a powerful and practical criterion for large-scale multiple testing problems. However, most FDR controlling approaches used in neuroimaging studies ignore the spatial dependence among the test statistics obtained from brain voxels and thus lose substantial power to effectively and accurately identify AD-related regions. This project aims to conduct a comparative study of existing FDR controlling approaches, in particular, recent spatial FDR methods, on AD's FDG-PET neuroimaging data to discover new and important AD-related brain regions that are missed by conventional FDR methods. We will use the FDG-PET data from the Alzheimer’s Disease Neuroimaging Initiative, and compare the CMRgl differences between the three disease-status groups, including AD patients, mild cognitive impairment patients, and cognitively normal subjects. |
| Investigator's Name: | Angie Gonzalez |
| Proposed Analysis: | Alzheimer’s disease (AD) is the most common type of dementia, accounting for approximately 60–80% of all cases. As a neurodegeneration biomarker, Fluorine-18 fluorodeoxyglucose positron emission tomography (FDG-PET) measures the cerebral metabolic rate of glucose (CMRgl) as a proxy of neural activity, and is extensively used in early diagnosis and monitoring progression of AD. The CMRgl difference between two disease-status groups can be investigated by testing the difference in the mean values of CMRgl at each brain image voxel, thereby becoming a high-dimensional multiple testing problem. The false discovery rate (FDR), a measure of type I error, provides a powerful and practical criterion for large-scale multiple testing problems. However, most FDR controlling approaches used in neuroimaging studies ignore the spatial dependence among the test statistics obtained from brain voxels and thus lose substantial power to effectively and accurately identify AD-related regions. This project aims to conduct a comparative study of existing FDR controlling approaches, in particular, recent spatial FDR methods, on AD's FDG-PET neuroimaging data to discover new and important AD-related brain regions that are missed by conventional FDR methods. We will use the FDG-PET data from the Alzheimer’s Disease Neuroimaging Initiative, and compare the CMRgl differences between the three disease-status groups, including AD patients, mild cognitive impairment patients, and cognitively normal subjects. |
