This repository contains the development for the paper "Does pre-training on brain-related tasks results in better deep-learning-based brain age biomarkers?", submitted to BRACIS'23 (citation soon). The slides presented in BRACIS are available in slides.pdf.
We use wandb to keep models and results.
The train_brainage.py and train_brats_models.py are the basis for reproducing our results.
Of course, they assume you already have the data ready for training.
The ADNI data we use are the standard split for ADNI1 (which is already a well-defined collection in loni's platform) and the images specified in data/raw/ADNI*_image_ids.csv.
Scripts src/preprocess.py and src/make_dataset.py do the job of getting the data ready for training.
With respect to BraTS pre-training, you must follow the steps of nnUNet to prepare the data.
The brain age prediction results can be seen with detail in our W&B report.
All statistical analysis mentioned in the paper come from statistical_analysis_results.csv.