Enhancing tumour microstructural quantification with machine learning and diffusion-relaxation MRI. Journal of Magnetic Resonance Imaging.
clinical2advanced is a straight-forward method for densly-sampled diffusion-relaxation MR images prediction from routine diffusion-only images. It also provides steps for advanced quantitative diffusion-relaxation metrics prediction from simple ADC estimations, despite the first approach is recommended.
run_me.py
: code to run the pipeline. Writerun_me.py -h
orrun_me.py --help
in your promt to print the help manualoptions.py
: class defining the mandatory and optional inputsdata_loader.py
: contains the functions to load and preprocess the datautils.py
: class containing multiple utils functions used troughtout the pipelinepatcher.py
: contains to functions to patch and unpatch the data if required by the usertrain_test.py
: contains the functions to train and/or test the different algorithmsmyNet.py
: class of multi-layer network.
This repository allows the prediction step of both approaches defined in Macarro C. et al. 2024. The fitting steps are not included here.
- maps-to-maps: diffusion-relaxation advanced quantitative parametric maps are predicted from simpler ADC measurements
- signal-to-signal: diffusion-relaxation signals are predicted from simpler diffusion-only signals
The code has been written in python v3.8.12 and the following packages are required:
numpy v1.20.3
nibabel v3.2.1
torch v1.11
scikit-learn v1.0.2
joblib v1.1.0
If you find our work useful, please consider citing our paper:
Macarro, C., Bernatowicz, K., Garcia-Ruiz, A. et al. (2024), Enhancing Tumor Microstructural Quantification With Machine Learning and Diffusion-Relaxation MRI. J Magn Reson Imaging. https://doi.org/10.1002/jmri.29484
This repository is distributed under the Attribution-NonCommercial-ShareAlike 4.0 International license (CC BY-NC-SA 4.0). Copyright (c) 2024, Fundació Privada Institut d’Investigació Oncològica de Vall d’Hebron (Vall d'Hebron Institute of Oncology (VHIO), Barcelona, Spain). All rights reserved. See licence here: LICENSE.txt