Deep learning framework for MRI reconstruction
We're excited to announce DIRECT v2.0.0, featuring several advancements and updates. Here's a snapshot of what's new:
Dive into the details below to see how DIRECT v2.0.0 can enhance your work.
RecurrentVarNet winner at MC-MRI challenge 2022, vSHARPNet winner at CMRxRecon challenge 2023) (#156, #180, #228, #271, #273)direct train and direct predict commands (#202)h5py (#174)ConjGradNet for improved imaging, IterDualNet (similar to JointICNet without sensitivity map optimization), ResNet as a new denoiser model, VarSplitNet for variable splitting optimization with deep learning (#228) and VSharpNet as presented in vSHARP: variable Splitting Half-quadratic ADMM algorithm for Reconstruction of inverse-Problems along with its 3D variant VSharpNet3D (#270, #273).EspiritCalibration transform via power-method algorithm, CropKspace, RandomFlip, RandomRotation., ComputePadding, ApplyPadding, ComputeImage , RenameKeys, ComputeScalingFactor.NMSE, NRMSE, NMAE, SobelGradL1Loss, SobelGradL2Loss, hfen_l1, hfen_l2, HFENLoss, HFENL1Loss, HFENL2Loss, snr, SNRLoss and SSIM3DLoss (#226, #262).Gaussian1DMaskFunc and Gaussian2DMaskFunc, respectively. Implemented using Cython (#230).VSharpNet3D, (#273).forward_method instead of _do_iteration. (#226)DictConfig from omegaconf using dict_flatten (#235, #250).DirectEnum as a base class for clean typing of options of modules such as transforms, etc (#228, #266).black reformatting (#241)This release was made possible by the hard work and dedication of our team and contributors:
Access detailed documentation for DIRECT v2.0.0 at our documentation site.
modulus_if_complex reinstated but requiring to set complex axisMRIEngine check if prediction is complex (complex axis=last) and apply the modulus if they are (#208).Full Changelog: https://github.com/NKI-AI/direct/compare/v1.0.2...v1.0.3
NormConv2dGRU) following the implementation of NormUnet2d (#176)SheppLoganProtonDataset
SheppLoganT1Dataset
SheppLoganT2Dataset
modulus_if_complex function removed, modulus needs to specify axis (#194)mypy and pylint fixes (#196)direct train, direct predict and python3 projects/predict_val.py to not necessarily require path to data as some datasets don't require it (e.g. SheppLogan Datasets) - build_dataset_from_input relies on **kwargs now. Refactored configs and docs to comply with the above. (#202)
Train command example:
direct train <experiment_directory> --num-gpus <number_of_gpus> --cfg <path_or_url_to_yaml_file> \ [--training-root <training_data_root> --validation-root <validation_data_root>] [--other-flags]
Normalize transform due to new version dependancy (#177)Full Changelog: https://github.com/NKI-AI/direct/compare/v1.0.1...v1.0.2
NMSE, NRMSE, NMAE, SobelGradL1Loss, SobelGradL2Loss) and also k-space losses (#226)ComputeZeroPadding: computes padding in k-space input (#226)ApplyZeroPadding: applies padding (#226)ComputeImage: computes image from k-space input (#226)RenameKeys: rename keys in input (#226)CropKSpace: transforms k-space to image domain, crops it, and backprojects it (#210)ApplyMask: applies sampling mask (#210)mri_models performs _do_iteration method, child engines perform forward_function which returns output_image and/or output_kspace (#226)torch.where output needed to be made contiguous to be inputted to fft, due to new torch version (#216)Full Changelog: https://github.com/NKI-AI/direct/compare/v1.0.3...v1.0.4
In v1.0.1 we mainly provide bug fixes (including a memory leak) and code quality improvements.
direct upload-to-bucket tool by @jonasteuwen in https://github.com/NKI-AI/direct/pull/167
h5py 3.6.0 to h5py 3.3.0
FastMRIDataset header reader function refactoreddo_iteration method. MRIModelEngine now includes all methods for MRI models.evaluate method of MRI Models has been rewritten.Full Changelog: https://github.com/NKI-AI/direct/compare/v1.0.0...v1.0.1
In the first stable release, we provide the implementation of baselines, including trained models on the publicly available datasets.
direct train and direct predict replace the corresponding python scripts (#151, #139)direct train CLI interface (Closes #109, #139)MRIReconstruction is merged with RIM (#113)state_dict if key is of the format __<>__ (#144 closes #143)Full Changelog: https://github.com/directgroup/direct/compare/v0.2...v1.0.0
Many new features have been added, of which most will likely have introduced breaking changes. Several performance issues have been addressed.
An improved version to the winning solution for the Calgary-Campinas challenge is also added to v0.2, including model weights.
model_zoo.md)train_rim.py by passing a directory.direct.utils.io.write_json serializes real-valued numpy and torch objects.direct.utils.str_to_class now supports partial argument parsing, e.g. fft2(centered=False) will be properly parsed
in the config.rim_engine.py.direct.data.samplers.BatchVolumeSampler returned wrong length which caused dropping of volumes during validation.In this version we provide necessary bugfixes for v0.1.1 and several improvements:
Black is now used as a code formatter, this had the consequence that certain parts became hard to read, so these were refactored to improve readability after applying black.
As you can expect from a pre-release, while we intend to keep it to a minimum, it is possible that things break, especially in the configuration files. In case you encounter one, please open an issue.
In v0.1.1:
--mixed-precision flag. If you have supporting hardware this can speed up training with more than 33% and reduce memory by more than 40%.In this release, we have significantly expanded upon v0.0.1 and: