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: | Loukas Ilias |
| Institution: | National Technical University of Athens |
| Department: | School of Electrical and Computer Engineering |
| Country: | |
| Proposed Analysis: | I am a PhD Candidate working towards the detection of complex brain disorders. Specifically, existing research initiatives use Convolutional Neural Networks (CNNs) or Long-Short Term Memory (LSTM) networks for categorizing groups into AD (Alzheimer’s Disease), MCI (Mild Cognitive Impairment), and HC (Healthy Control). Despite the success of transformer-based networks [1] in many fields, little work has been done towards alzheimer’s disease detection. Thus, we aim to propose advanced transformer networks, which will achieve comparable or new state-of-the-art results. One further future plan is to propose multimodal models. Existing state-of-the-art approaches concatenate the representations obtained by the different modalities, thus treating equally each modality. We aim to tackle this limitation by introducing methods for fusing the different modalities effectively. Examples constitute the gated multimodal unit [2], co-attention mechanism [3] , and many more. In addition, neural networks are considered black-boxes. In order to tackle this limitation, we aim to propose explainability techniques, including LIME, Shapley Values, Integrated Gradients, etc. Concurrently, we aim to propose interpretable models without requiring the need for a post-hoc explainability analysis. I have published several papers regarding the detection of AD patients from spontaneous speech. [1] Vaswani, Ashish, et al. "Attention is all you need." Advances in neural information processing systems 30 (2017). [2] Arevalo, John, et al. "Gated multimodal networks." Neural Computing and Applications 32.14 (2020): 10209-10228. [3] Lu, Jiasen, et al. "Hierarchical question-image co-attention for visual question answering." Advances in neural information processing systems 29 (2016). |
| Additional Investigators | |
| Investigator's Name: | Dimitris Askounis |
| Proposed Analysis: | The data will be used by a PhD candidate working towards the detection of complex brain disorders. Specifically, existing research initiatives use Convolutional Neural Networks (CNNs) or Long-Short Term Memory (LSTM) networks for categorizing groups into AD (Alzheimer’s Disease), MCI (Mild Cognitive Impairment), and HC (Healthy Control). Despite the success of transformer-based networks [1] in many fields, little work has been done towards alzheimer’s disease detection. Thus, we aim to propose advanced transformer networks, which will achieve comparable or new state-of-the-art results. One further future plan is to propose multimodal models. Existing state-of-the-art approaches concatenate the representations obtained by the different modalities, thus treating equally each modality. We aim to tackle this limitation by introducing methods for fusing the different modalities effectively. Examples constitute the gated multimodal unit [2], co-attention mechanism [3] , and many more. In addition, neural networks are considered black-boxes. In order to tackle this limitation, we aim to propose explainability techniques, including LIME, Shapley Values, Integrated Gradients, etc. Concurrently, we aim to propose interpretable models without requiring the need for a post-hoc explainability analysis. [1] Vaswani, Ashish, et al. "Attention is all you need." Advances in neural information processing systems 30 (2017). [2] Arevalo, John, et al. "Gated multimodal networks." Neural Computing and Applications 32.14 (2020): 10209-10228. [3] Lu, Jiasen, et al. "Hierarchical question-image co-attention for visual question answering." Advances in neural information processing systems 29 (2016). |
