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: | Senuri De Silva |
| Institution: | University of Moratuwa |
| Department: | Computer Science and Engineering |
| Country: | |
| Proposed Analysis: | Adults with Alzheimer's Disease (AD) have the tendency to develop crucial brain damage which will ultimately result in death. Hence the early detection of AD is predominant to start off the medication to reduce the progression and catastrophic results. Currently, the studies have focused on studying the anatomical differences in the brain using the structural MRI data where functional neuroimaging would be adding a new perspective to AD’s pathology discovery. Hence to assist the current usage of structural MRI data along with the PET images, functional MRI has the potential to develop a novel dimension and build upon. Therefore, there is a major requirement to generate an objective biological tool that is capable of classifying AD and incorporating both structural and functional aspects as different modalities in the classification process. The proposed solution contains six steps including data acquisition, preprocessing, feature extraction, machine learning model derivation, and generating the ensemble model. As this research supports three modalities of data containing functional MRI, structural MRI and PET scanning data, preprocessing techniques should be applied to each of these modalities. The fMRI preprocessing may include a pipeline of slice-time correction, motion correction, normalization where the structural MRI also may include coregistration, normalization, and the PET images will be subjected to a preprocessing pipeline consists of coregistration and segmentation. Then the pre-processed data will be ready for the feature extraction step where the fALFF, ReHo features are to be considered for fMRI data and regions of interest (ROI) to be considered as per the structural MRI and PET images. After the feature extraction step, the data is subjected to the application of machine learning and deep learning algorithms such as convolutional neural networks, deep neural nets, etc. Each modality of data will generate a separate model and after evaluating and tuning the hyperparameters, an ensemble model can be obtained to support the AD classification from healthy subjects and can be even extended to identify between different stages of AD. |
| Additional Investigators | |
| Investigator's Name: | Sanuwani Dayarathna |
| Proposed Analysis: | Adults with Alzheimer's Disease (AD) have the tendency to develop crucial brain damage which will ultimately result in death. Hence the early detection of AD is predominant to start off the medication to reduce the progression and catastrophic results. Currently, the studies have focused on studying the anatomical differences in the brain using the structural MRI data where functional neuroimaging would be adding a new perspective to AD’s pathology discovery. Hence to assist the current usage of structural MRI data along with the PET images, functional MRI has the potential to develop a novel dimension and build upon. Therefore, there is a major requirement to generate an objective biological tool that is capable of classifying AD and incorporating both structural and functional aspects as different modalities in the classification process. The proposed solution contains six steps including data acquisition, preprocessing, feature extraction, machine learning model derivation, and generating the ensemble model. As this research supports three modalities of data containing functional MRI, structural MRI and PET scanning data, preprocessing techniques should be applied to each of these modalities. The fMRI preprocessing may include a pipeline of slice-time correction, motion correction, normalization where the sMRI also may include coregistration, normalization, and the PET images will be subjected to a preprocessing pipeline consists of coregistration and segmentation. Then the pre-processed data will be ready for the feature extraction step where the fALFF, ReHo features are to be considered for fMRI data and regions of interest (ROI) to be considered as per the sMRI and PET images. After the feature extraction step, the data is subjected to the application of machine learning and deep learning algorithms such as convolutional neural networks, deep neural nets, etc. Each modality of data will generate a separate model and after evaluating and tuning the hyperparameters, an ensemble model can be obtained to support the AD classification from healthy subjects and can be even extended to identify between different stages of AD. |
| Investigator's Name: | Dulani Meedeniya |
| Proposed Analysis: | Adults with Alzheimer's Disease (AD) have the tendency to develop crucial brain damage which will ultimately result in death. Hence the early detection of AD is predominant to start off the medication to reduce the progression and catastrophic results. Currently, the studies have focused on studying the anatomical differences in the brain using the structural MRI data where functional neuroimaging would be adding a new perspective to AD’s pathology discovery. Hence to assist the current usage of structural MRI data along with the PET images, functional MRI has the potential to develop a novel dimension and build upon. Therefore, there is a major requirement to generate an objective biological tool that is capable of classifying AD and incorporating both structural and functional aspects as different modalities in the classification process. The proposed solution contains six steps including data acquisition, preprocessing, feature extraction, machine learning model derivation, and generating the ensemble model. As this research supports three modalities of data containing functional MRI, structural MRI and PET scanning data, preprocessing techniques should be applied to each of these modalities. The fMRI preprocessing may include a pipeline of slice-time correction, motion correction, normalization where the sMRI also may include coregistration, normalization, and the PET images will be subjected to a preprocessing pipeline consists of coregistration and segmentation. Then the pre-processed data will be ready for the feature extraction step where the fALFF, ReHo features are to be considered for fMRI data and regions of interest (ROI) to be considered as per the sMRI and PET images. After the feature extraction step, the data is subjected to the application of machine learning and deep learning algorithms such as convolutional neural networks, deep neural nets, etc. Each modality of data will generate a separate model and after evaluating and tuning the hyperparameters, an ensemble model can be obtained to support the AD classification from healthy subjects and can be even extended to identify between different stages of AD. |
| Investigator's Name: | Mawli De Silva |
| Proposed Analysis: | Adults with Alzheimer's Disease (AD) have the tendency to develop crucial brain damage which will ultimately result in death. Hence the early detection of AD is predominant to start off the medication to reduce the progression and catastrophic results. Currently, the studies have focused on studying the anatomical differences in the brain using the structural MRI data where functional neuroimaging would be adding a new perspective to AD’s pathology discovery. Hence to assist the current usage of structural MRI data along with the PET images, functional MRI has the potential to develop a novel dimension and build upon. Therefore, there is a major requirement to generate an objective biological tool that is capable of classifying AD and incorporating both structural and functional aspects as different modalities in the classification process. The proposed solution contains six steps including data acquisition, preprocessing, feature extraction, machine learning model derivation, and generating the ensemble model. As this research supports three modalities of data containing functional MRI, structural MRI and PET scanning data, preprocessing techniques should be applied to each of these modalities. The fMRI preprocessing may include a pipeline of slice-time correction, motion correction, normalization where the sMRI also may include coregistration, normalization, and the PET images will be subjected to a preprocessing pipeline consists of coregistration and segmentation. Then the pre-processed data will be ready for the feature extraction step where the fALFF, ReHo features are to be considered for fMRI data and regions of interest (ROI) to be considered as per the sMRI and PET images. After the feature extraction step, the data is subjected to the application of machine learning and deep learning algorithms such as convolutional neural networks, deep neural nets, etc. Each modality of data will generate a separate model and after evaluating and tuning the hyperparameters, an ensemble model can be obtained to support the AD classification from healthy subjects and can be even extended to identify between different stages of AD. |
