We release Hybrid-FAZ Alzheimer's diagnosis code.

Contributors: Chae-yeon Lim, Kyungsu Kim

Detailed instructions for diagnosis Alzheimer's are as follows.

Pytorch implementation based on radiomics multi-feature based machine learning using FAZ segmentation image

Using Paper Dataset

Proposed: Hybrid-FAZ/Holdout_proposed_AI/*

baseline: Hybrid-FAZ/Holdout_baseline_manual/*

Required libraries for training/inference are detailed in requirements.txt

pip install -r requirements.txt

Conducting training through nnUNet using a public data set

Public dataset: https://zenodo.org/record/5075563

AI-based segmentation result

Radiomics multi-feature based machine learning diagnosis

Baseline code: path/to/Hybrid-FAZ/FAZ_code/FAZ_Holdout_test_baseline.ipynb

Proposed code: path/to/Hybrid-FAZ/FAZ_code/FAZ_Holdout_test_propose.ipynb

thanks to MIC-DKFZ for sharing the segmentation nnUNet code.

nnUNet github: https://github.com/MIC-DKFZ/nnUNet