Non-Cartesian-SSDU-MRI

KO | EN

gitlite — search

#java #csharp #typescript #monitoring #ai #http #automation #large-language-models #github #pdf

Non-Cartesian-SSDU-MRI

(★ 9)

No description available.

File Explorer

ZIP Download

- all_ones_mask.mat
- traj_figure.png
- trj_kxy_full.mat
- trj_kxy_single.mat
- tsnr_figure.png
- Tuke_01.mat
- Tukey_32.mat
- Tukey_64.mat
- - ._.DS_Store
  - ._demo_CC_speedup.m
  - ._demo_ECC.m
  - ._demo_GCC.m
  - ._setPath.m
  - .demo_GCC.m.swp
  - .DS_Store
  - brain_8ch.mat
  - demo_CC_speedup.m
  - demo_ECC.m
  - demo_ESPIRiT_L1_recon.m
  - demo_ESPIRiT_maps.m
  - demo_ESPIRiT_recon.m
  - demo_GCC.m
  - demo_l1_spirit_pocs.m
  - demo_nuESPIRiT.m
  - demo_nuSPIRiT.m
  - demo_SAKE.m
  - demo_spirit_cg.m
  - espirit_generator.m
  - espirit_sigpy.py
  - Espirit_Torch.py
  - Esprit.py
  - LICENSE.txt
  - make.m
  - README.m
  - setPath.m
- - arg_get.m
  - arg_pair.m
  - beeper.m
  - blank_unmash.m
  - bspline_1d_coef.m
  - bspline_1d_interp.m
  - bspline_1d_synth.m
  - caller_name.m
  - cg_demo1.m
  - coils.mat
  - coils_1_new.mat
  - coils_2_new.mat
  - coils_70_3.mat
  - coils_70_9.mat
  - coils_95_9.mat
  - col.m
  - complexify.m
  - cpu.m
  - data.mat
  - dcf.mat
  - deg2rad.m
  - demo.m
  - demo_0_4.m
  - demo_0_4_fixed.m
  - demo_0_4_fixed_figure.m
  - demo_0_4_fixed_more_sbj.m
  - demo_0_4_fixed_more_sbj_7.m
  - demo_0_4_fixed_more_sbj_7_plot_dcf_M.m
  - demo_0_4_fixed_more_sbj_7_plot_trj_dcf_M.m
  - demo_0_4_fixed_more_sbj_7_plot_trj_dcf_M.m~
  - demo_0_4_fixed_more_sbj_output_0.m
  - demo_0_4_fixed_more_sbj_output_paper.m
  - demo_0_4_fixed_more_sbj_output_paper_80.m
  - demo_0_4_fixed_more_sbj_output_paper_80_from_full.m
  - demo_0_4_fixed_more_sbj_output_paper_from_full.m
  - demo_0_4_fixed_more_sbj_paper.m
  - demo_0_4_fixed_more_sbj_paper_half.m
  - demo_0_4_fixed_more_sbj_paper_last3.m
  - demo_0_4_fixed_more_sbj_paper_less.m
  - demo_0_4_more.m
  - demo_0_4_more_fixed.m
  - demo_0_5_ssdu.m
  - demo_531_new_data_old.m
  - demo_hy.m
  - demo_new_data.m
  - demo_singleshot_newdata.m
  - demo_spiral.m
  - demo_spiral_another_version.m
  - demo_spiral_new_2d_data.m
  - demo_spiral_new_2d_data_10x.m
  - demo_spiral_new_2d_data_10x_fixed_actual.m
  - demo_spiral_new_2d_data_10x_fixed_actual_large.m
  - demo_spiral_new_2d_data_10x_fixed_actual_large_new.m
  - demo_spiral_new_2d_data_10x_fixed_actual_large_new1027.m
  - demo_spiral_new_2d_data_10x_fixed_actual_large_new1027_paper.m
  - demo_spiral_new_2d_data_10x_fixed_actual_large_new1027_paper_80.m
  - demo_spiral_new_2d_data_10x_fixed_actual_large_R10.m
  - demo_spiral_new_2d_data_full.m
  - demo_spiral_new_2d_data_full_fixed.m
  - demo_spiral_new_2d_data_full_fixed_large.m
  - demo_spiral_new_2d_data_full_fixed_large_new.m
  - demo_spiral_new_2d_data_full_fixed_large_new.m~
  - demo_spiral_new_2d_data_full_fixed_large_new1027.m
  - demo_spiral_new_2d_data_full_fixed_large_new1027_paper.m
  - demo_spiral_new_2d_data_full_fixed_large_new1027_paper_80.m
  - demo_spiral_r5.m
  - demo_spiral_singleshot.m
  - demo_spiral_singleshot_for_cg.m
  - demo_spiral_singleshot_ssdu.m
  - demo_spiral_singleshot_ssdu_new.m
  - demo_spiral_singleshot_ssdu_other.m
  - demo_spiral_singleshot_ssdu_simple.m
  - demo_spiral_singleshot_ssdu_simple_new_data.m
  - demo_spiral_singleshot_ssdu_simple_newdata.m
  - demo_spiral_singleshot_ssdu_simple_newdata725.m
  - demo_spiral_singleshot_ssdu_simple_newdata_5mask_again.m
  - demo_spiral_singleshot_test_cg.m
  - demo_spiral_singleshot_test_cg2.m
  - demo_spiral_singleshot_test_cg_lambda.m
  - demo_spiral_under_10_first.m
  - demo_spiral_under_10_first_newdata.m
  - demo_spiral_under_10_first_newdata_fixed_large.m
  - demo_spiral_x10_newdata.m
  - demo_spiral_x10_newdata_fixed.m
  - demo_testEHWY.m
  - demo_testEHWY_for_ssdu.m
  - demo_under.m
  - demo_under_smaller.m
  - demo_W.m
  - demo_W_spiral.m
  - demo_W_spiral_full.m
  - demo_W_spiral_full_531_modified_new_data.m
  - demo_W_spiral_full_618_modified_new_data.m
  - demo_W_spiral_full_newdata.m
  - demo_W_spiral_full_newdata_919.m
  - demo_W_spiral_R10_newdata.m
  - demo_W_spiral_R10_newdata.m~
  - demo_W_spiral_ssdu_0_5m
  - demo_W_spiral_ssdu_x10.m
  - demo_W_spiral_ssdu_x10_0_5.m
  - demo_W_spiral_ssdu_x10_new.m
  - demo_W_spiral_ssdu_x10_new_data_old.m
  - demo_W_spiral_ssdu_x10_new_x.m
  - demo_W_spiral_ssdu_x10_new_x_show_EHY.m
  - demo_W_spiral_ssdu_x10_new_x_show_EHY_0_5.m
  - demo_W_spiral_ssdu_x10_new_x_show_EHY_0_5_725_10mask.m
  - demo_W_spiral_ssdu_x10_new_x_show_EHY_0_5_725_5mask.m
  - demo_W_spiral_ssdu_x10_new_x_show_EHY_other_dcf.m
  - demo_W_spiral_x.m
  - demo_W_spiral_x10_newdata.m
  - demo_W_spiral_x_multiply_by_2.m
  - demo_W_spiral_x_multiply_original.m
  - demo_W_spiral_x_test_scaling.m
  - demo_W_spiral_x_test_scaling_full.m
  - demo_W_spiral_x_test_scaling_same_dcf.m
  - demo_W_spiral_x_test_scaling_same_dcf_how_close.m
  - demo_W_spiral_x_test_scaling_single.m
  - demo_W_ssdu_725_5_mask_lossy.m
  - detex.m
  - dft_sym_check.m
  - diffc.m
  - difff.m
  - dims_same.m
  - div0.m
  - double6.m
  - doubles.m
  - downsample1.m
  - downsample2.m
  - downsample3.m
  - dsingle.m
  - dtft.m
  - dtft1.m
  - dtft2.m
  - dtft2_adj.m
  - dtft_adj.m
  - dtft_mex.m
  - dtft_test.m
  - embed.m
  - equivs.m
  - example_nufft1_reverse.m
  - example_nufft_antenna.m
  - fail.m
  - fftn_fast.m
  - fig_nufft_emax.m
  - fig_taper_cos3.m
  - filtmat.m
  - fld_read.m
  - fld_write.m
  - flipdims.m
  - fractional_delay.m
  - fwhm.m
  - fwhm1.m
  - fwhm2.m
  - fwhm_match.m
  - gaussian_kernel.m
  - get_image.m
  - get_image_fmri.m
  - group2d.m
  - Gsparse.m
  - has_aspire.m
  - has_mex_jf.m
  - highrate_centers.m
  - hist_bin_int.m
  - hist_equal.m
  - ifft_sym.m
  - ifftn_fast.m
  - imax.m
  - img.mat
  - imin.m
  - interp1_jump.m
  - interp1_lagrange.m
  - interp1x.m
  - interp_table_test.m
  - is_pre_v7.m
  - isfreemat.m
  - isvar.m
  - jf.m
  - jf_assert.m
  - jf_dcm_write.m
  - jf_equal.m
  - jf_histn.m
  - jf_protected_names.m
  - jf_whos_nan.m
  - jinc.m
  - kaiser,m=0.mat
  - kaiser,m=2.mat
  - kaiser_bessel.m
  - kaiser_bessel_ft.m
  - kaiser_bessel_xray.m
  - kb,notes,matej
  - kde_pmf1.m
  - kde_pmf2.m
  - kde_pmf_width.m
  - ksp.mat
  - L1reg.m
  - L2reg.m
  - lloyd_max_hist.m
  - load_ascii_skip_header.m
  - M_1.m
  - M_2.m
  - M_3.m
  - M_4.m
  - M_5.m
  - M_6.m
  - M_full_120.m
  - M_full_80.m
  - M_ssdu.mat
  - M_test.m
  - M_test_80.m
  - mag_angle_real.m
  - mask_paper_1_1.mat
  - mask_paper_1_10.mat
  - mask_paper_1_2.mat
  - mask_paper_1_3.mat
  - mask_paper_1_4.mat
  - mask_paper_1_5.mat
  - mask_paper_1_6.mat
  - mask_paper_1_7.mat
  - mask_paper_1_8.mat
  - mask_paper_1_9.mat
  - mask_paper_2_1.mat
  - mask_paper_2_10.mat
  - mask_paper_2_2.mat
  - mask_paper_2_3.mat
  - mask_paper_2_4.mat
  - mask_paper_2_5.mat
  - mask_paper_2_6.mat
  - mask_paper_2_7.mat
  - mask_paper_2_8.mat
  - mask_paper_2_9.mat
  - mask_paper_3_1.mat
  - mask_paper_3_10.mat
  - mask_paper_3_2.mat
  - mask_paper_3_3.mat
  - mask_paper_3_4.mat
  - mask_paper_3_5.mat
  - mask_paper_3_6.mat
  - mask_paper_3_7.mat
  - mask_paper_3_8.mat
  - mask_paper_3_9.mat
  - mask_paper_4_1.mat
  - mask_paper_4_10.mat
  - mask_paper_4_2.mat
  - mask_paper_4_3.mat
  - mask_paper_4_4.mat
  - mask_paper_4_5.mat
  - mask_paper_4_6.mat
  - mask_paper_4_7.mat
  - mask_paper_4_8.mat
  - mask_paper_4_9.mat
  - mask_paper_5_1.mat
  - mask_paper_5_10.mat
  - mask_paper_5_2.mat
  - mask_paper_5_3.mat
  - mask_paper_5_4.mat
  - mask_paper_5_5.mat
  - mask_paper_5_6.mat
  - mask_paper_5_7.mat
  - mask_paper_5_8.mat
  - mask_paper_5_9.mat
  - mask_paper_6_1.mat
  - mask_paper_6_10.mat
  - mask_paper_6_2.mat
  - mask_paper_6_3.mat
  - mask_paper_6_4.mat
  - mask_paper_6_5.mat
  - mask_paper_6_6.mat
  - mask_paper_6_7.mat
  - mask_paper_6_8.mat
  - mask_paper_6_9.mat
  - masker.m
  - mat_read.m
  - max_percent_diff.m
  - min_cos_quad.m
  - minmax.m
  - mod0.m
  - my_dcf.m
  - nans.m
  - ncol.m
  - ndgrid_jf.m
  - newfft.m
  - newfft_approx_adj.m
  - newfft_approx_for.m
  - newfft_exact_adj.m
  - newfft_exact_for.m
  - newfft_exact_p.m
  - newfft_table_init.m
  - nrms.m
  - nrow.m
  - nufft.m
  - nufft1_build.m
  - nufft1_emax.m
  - nufft1_err_mm.m
  - nufft1_error.m
  - nufft1_kernel.m
  - nufft1_time.m
  - nufft2_err_mm.m
  - nufft_adj - Copy.m
  - nufft_adj.m
  - nufft_alpha.m
  - nufft_alpha_kb_fit.m
  - nufft_best_alpha.m
  - nufft_best_gauss.m
  - nufft_coef.m
  - nufft_diric.m
  - nufft_gauss.m
  - nufft_gauss2.mat
  - nufft_init.m
  - nufft_interp_zn.m
  - nufft_offset.m
  - nufft_r.m
  - nufft_samples.m
  - nufft_scale.m
  - nufft_scale_kb.m
  - nufft_sinc.m
  - nufft_T.m
  - nufft_table_adj.m
  - nufft_table_change.m
  - nufft_table_init.m
  - nufft_table_interp.m
  - nufft_table_test.m
  - nufft_tune_gauss.m
  - nufft_tune_kaiser.m
  - num2list.m
  - Omega.png
  - os_run.m
  - outer_sum.m
  - pad_into_center.m
  - padn.m
  - path_find_dir.m
  - poisson.m
  - poisson0.m
  - poisson1.m
  - poisson1.orig
  - poisson2.m
  - poisson2unif.m
  - poly_string.m
  - pr.m
  - printf.m
  - printm.m
  - printv.m
  - prompt.m
  - rad2deg.m
  - read.m
  - reale.m
  - rect.m
  - Ref_omega.png
  - remove_spaces.m
  - repeat_slice.m
  - repout.m
  - reshapee.m
  - reshaper.m
  - rms.m
  - run_mfile_local.m
  - see_coil_7_31.m
  - show_M.m
  - sinc_periodic.m
  - single_ws.m
  - sino_mash.m
  - spdiag.m
  - stackpick.m
  - stackup.m
  - streq.m
  - strreps.m
  - strum_test.m
  - subv2ind.m
  - swapdim.m
  - test_all.m
  - test_all_nufft.m
  - test_all_util.m
  - test_dir.m
  - ticker.m
  - traj.mat
  - truncate_precision.m
  - unpadn.m
  - upsample_rep.m
  - vararg_pair.m
  - vcorrcoef.m
  - W_0_4.m
  - W_0_4_more.m
  - W_10_mask_more.m
  - warn.m
  - ZcCGSENSE3D_SingleEcho.m
  - ZcGetSpiralCoils_2D.m
  - zero_tiny_negative.m
- A0_DCF.py
- A0_MaskGen.m
- A1_FastOperator.m
- A2_LabelCoilGen.m
- A3_EHyGen.m
- A4_EHyGen_TestfMRI.m
- ReadMe.md
- DataLoader.py
- DataLoader_Test.py
- DC.py
- ResNet.py
- test.py
- train.py
- Unrolled_Network.py
README.md
requirements.txt

# Use via CDN

jsDelivr

jsDelivr serves any public GitHub repository as a CDN with zero setup. Pick a version and a file to get a ready-to-paste link and snippet.

Version

File

Link

Example

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

## Abstract

Physics-driven deep learning (PD-DL) methods have shown strong promise for accelerated MRI reconstruction by combining learned image priors with the physical MRI forward model. However, conventional PD-DL pipelines can struggle when the target images contain large phase variations, since these phase inconsistencies may degrade data consistency and reconstruction quality. This issue is especially relevant in multi-echo fMRI, where substantial phase differences can appear across echo images.

To address this challenge, we propose a low-passed filter phase-corrected physics-driven deep learning method, referred to as LPF-based PC-PDDL, that explicitly models and corrects phase inconsistencies within the unrolled reconstruction process.

Experiments on prospectively undersampled non-Cartesian multi-echo fMRI data show that the low-pass filtering strategy improves upon standard non-phase-corrected PD-DL despite its simplicity. These results suggest that explicit phase correction can improve non-Cartesian multi-echo fMRI reconstruction, while also highlighting the importance of robust and generalizable phase estimation within unrolled PD-DL frameworks.

```text
Non-Cartesian-SSDU-MRI/
└── Data/
   ├── Tuke_01.mat
   ├── Tukey_32.mat
   ├── Tukey_64.mat
   ├── all_ones_mask.mat
   ├── trj_kxy_full.mat
   └── trj_kxy_single.mat

└── Preprocess/
   ├── Espirit/
   ├── nufft/
   ├── A0_DCF.py
   ├── A0_MaskGen.m
   ├── A1_FastOperator.m
   ├── A2_LabelCoilGen.m
   ├── A3_EHyGen.m
   ├── A4_EHyGen_TestfMRI.m
   └── README.md

└── Training/
   ├── DC.py
   ├── DataLoader.py
   ├── DataLoader_Test.py
   ├── ResNet.py
   ├── Unrolled_Network.py
   ├── train.py
   └── test.py
   
└── requirements.txt
└── README.md
```

## Quick Start
**Note:** This code was tested with `torch==2.2.1+cu121`.

### 1. Clone this repository

```bash
git clone https://github.com/MahdiSaberii/Non-Cartesian-SSDU-MRI.git
cd Non-Cartesian-SSDU-MRI
```

### 2. Create and activate conda environment

```bash
conda create -n non_cartesian_ssdu_mri python=3.10 -y
conda activate non_cartesian_ssdu_mri
```

### 3. Install PyTorch

```bash
pip install torch==2.2.1+cu121 --index-url https://download.pytorch.org/whl/cu121
```

### 4. Install remaining requirements

```bash
pip install -r requirements.txt
```

## Training and Testing

After completing the preprocessing pipeline, the prepared data can be used for multi-mask SSDU training. The preprocessing step should generate the required NUFFT operators, coil sensitivity maps, density compensation files, and SSDU training/validation masks.

Before starting training, set the desired number of training masks in `train.py`. This controls how many SSDU mask splits are used during multi-mask self-supervised learning.

To train the reconstruction network, run:

```bash
python train.py
```

After training is completed, the trained model can be tested on the full fMRI scan. Specify the path to the saved model weights in `test.py`, then run:

```bash
python test.py
```

The testing script loads the trained model weights and reconstructs the whole fMRI scan using the selected checkpoint.

## 📝 BibTeX

If you find this repository useful in your research, please consider citing our works:

```bibtex
@inproceedings{saberi2026phase,
  title={Phase-Correction Strategies for Physics-Driven Deep Learning Reconstruction of Accelerated Non-Cartesian Multi-Echo fMRI},
  author={Saberi, Mahdi and Yu, Zidan and Rettenmeier, Christoph and Stenger, Andrew and Ak{\c{c}}akaya, Mehmet},
  booktitle={2026 IEEE 23rd International Symposium on Biomedical Imaging (ISBI)},
  pages={1--4},
  year={2026},
  organization={IEEE}
}

@inproceedings{gu2024non,
  title={Non-cartesian self-supervised physics-driven deep learning reconstruction for highly-accelerated multi-echo spiral fMRI},
  author={Gu, Hongyi and Zhang, Chi and Yu, Zidan and Rettenmeier, Christoph and Stenger, V Andrew and Ak{\c{c}}akaya, Mehmet},
  booktitle={2024 IEEE International Symposium on Biomedical Imaging (ISBI)},
  pages={1--5},
  year={2024},
  organization={IEEE}
}

Non-Cartesian-SSDU-MRI

Owner Repos

File Explorer

# Use via CDN

Was this content helpful?

Recent Feedback

Non-Cartesian-SSDU-MRI

Owner Repos

File Explorer

# Use via CDN

Was this content helpful?

Recent Feedback

Download README