UPDATE:

• 03/2024: Published on Scientific Reports! Check the paper here.

Multimodal approach for AD Detection

This repository contains notebooks for the experiments conducted for testing a multimodal approach for assessing AD from PET and MRI. The notebooks are separated in:

• Dataset_MRI.ipynb, containing the code used to process the raw MRI data to create the final dataset.
• Dataset_PET.ipynb, containing the code used to process the raw PET data to create the final dataset.
• Training.ipynb, which contains the code used for training and testing the various models.
• Heatmaps.ipynb, which contains the code used for generating heatmaps based on the GradCAM algorithm, using the plugin tf-keras-vis.
• Exploration.ipynb, which contains the code used for the evaluation of the heatmaps in order to rank the brain's zones.
• images/3D Brain Plot.ipynb, which contains the code used to generate the figures available in the paper.