Comparison of Techniques for the Automatic Diagnosis of Alzheimer’s Disease

Published: 17-02-2020| Version 2 | DOI: 10.17632/d5x4mcjygf.2
Óscar Darias Plasencia


This repository contains all the code notebooks used for model implementation in the research "Comparison of Techniques for the Automatic Diagnosis of Alzheimer's Disease". The data used for training and testing could not be shared due to the fact that it is property of the Alzheimer's Disease Neuroimaging Initiative (ADNI), and the authors were allowed to access it only for the purpose of this research. The code is written in a series of separate Python Notebooks created and managed using Google Colaboratory. The files are being shared in this repository with the sole purpose of allowing further understanding of the results explained in the research paper. Reproducibility would be possible only if the interested individuals were able to access ADNI data.


Steps to reproduce

The main purpose of sharing these files is not reproducibility. However, steps could be reproduced if a user had access to ADNI MRI image data. If the user is in possession of these data, or really any other MRI data, it would only be necessary to follow the steps described on each notebook. The correct order of execution would be: 1. Image Preprocessing 2. Building TFRecords DB 3. Using ResNet3D + Fine-tuning of InceptionV3