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: | Aasa Feragen |
Institution: | Technical University of Denmark |
Department: | DTU Compute |
Country: | |
Proposed Analysis: | Analyze whether state-of-the-art algorithms for computed aided diagnosis of Alzheimers disease or MCI trained on the ADNI dataset, has a demographic bias in its performance when tested, with respect to groups defined by geography (AIBL, and we will also seek access to other WW ADNI datasets), gender, age and war veteran history (ADNIDOD). |
Additional Investigators | |
Investigator's Name: | Melanie Ganz |
Proposed Analysis: | Analyze whether state-of-the-art algorithms for computed aided diagnosis of Alzheimers disease or MCI trained on the ADNI dataset, has a demographic bias in its performance when tested, with respect to groups defined by geography (AIBL, and we will also seek access to other WW ADNI datasets), gender, age and war veteran history (ADNIDOD). |
Investigator's Name: | Maria Luise da Costa Zemsch |
Proposed Analysis: | Analyze whether state-of-the-art algorithms for computed aided diagnosis of Alzheimers disease or MCI trained on the ADNI dataset, has a demographic bias in its performance when tested, with respect to groups defined by geography (AIBL, and we will also seek access to other WW ADNI datasets), gender, age and war veteran history (ADNIDOD). Luise will carry out part of the analysis as part of her MSc thesis work. |
Investigator's Name: | Camilla Kergel Pedersen |
Proposed Analysis: | Analyze whether state-of-the-art algorithms for computed aided diagnosis of Alzheimers disease or MCI trained on the ADNI dataset, has a demographic bias in its performance when tested, with respect to groups defined by geography (AIBL, and we will also seek access to other WW ADNI datasets), gender, age and war veteran history (ADNIDOD). Camilla will carry out part of the analysis as part of her MSc thesis work. |
Investigator's Name: | Emily Beaman |
Proposed Analysis: | Analyze whether state-of-the-art algorithms for computed aided diagnosis of Alzheimers disease or MCI trained on the ADNI dataset, has a demographic bias in its performance when tested, with respect to groups defined by geography (AIBL, and we will also seek access to other WW ADNI datasets), gender, age and war veteran history (ADNIDOD). Emily is a research assistant associated with the project. |
Investigator's Name: | Eike Petersen |
Proposed Analysis: | Analyze gender bias in algorithms trained to predict Alzheimer diagnosis from MRI image features. |
Investigator's Name: | Oskar Christensen |
Proposed Analysis: | Train an image-to-image GAN that translates between 1.5T and 3T scans to use as augmentation for training diagnostic deep learning models from MRI images. The goal is to avoid bias by magnetic field strength, as we are currently observing such a bias in our models. |
Investigator's Name: | Anders Henriksen |
Proposed Analysis: | Train an image-to-image GAN that translates between 1.5T and 3T scans to use as augmentation for training diagnostic deep learning models from MRI images. The goal is to avoid bias by magnetic field strength, as we are currently observing such a bias in our models. |
Investigator's Name: | Elisabeth Zinck |
Proposed Analysis: | Develop a fairness barometer which tests predictive algorithms for their alignment with various definitions of algorithmic fairness. This will be tested on a range of datasets, where one consists of MRI features from ADNI subjects, used to train a diagnostic classifier. |
Investigator's Name: | Caroline Fuglsang Damgaard |
Proposed Analysis: | Develop a fairness barometer which tests predictive algorithms for their alignment with various definitions of algorithmic fairness. This will be tested on a range of datasets, where one consists of MRI features from ADNI subjects, used to train a diagnostic classifier. |