This project was done by jeongahblairlee and nigelhartm for the module Bioimage Analysis and Extended Phenotyping (Msc. Bioinformatics) in one week. As a dataset we used https://www.kaggle.com/datasets/sachinkumar413/alzheimer-mri-dataset. From there we tried to improve the prediction and implemented a SVM approach as well. Our main focus was in trying to understand the underlaying approaches/data needed to do a prediction of Alzheimer's disease types (Mild Demented, Moderate Demented, Non Demented, Very Mild Demented) based on brain MRI pictures. Fruther description and the presentation about our results.
We used Google colab to run our training and prediction, thats where our main.ipynb is coming from. Our best result we achieved by using a SVM, which we saved as a pre-trained model and provide here as well (SOON). This provided us an accuracy of 0.9825 in predicting the correct stage of Alzheimer's disease.
This we already tried at the CNN approach but didn't get good results. It is planned to do it on the SVM approach.
- optimzing model (open)
- create webpage for project (done) -> http://nigelhartm.github.io/alzheimer_disease_stages_prediction
- implement a webpage with tensorflow for javascript (open)
https://www.kaggle.com/datasets/jboysen/mri-and-alzheimers
https://ieeexplore.ieee.org/document/9521165
https://catalog.data.gov/dataset/alzheimers-disease-and-healthy-aging-data
https://www.nature.com/articles/s41598-020-79243-9
https://www.kaggle.com/code/psycon/brain-mri-image-alzheimer-classifier/notebook