ReMiDi - Reconstruction of Microstructure using Diffusion Magnetic Resonance Imaging

ReMiDi is a framework for reconstructing complex microstructural geometries from a diffusion MRI (dMRI) signal, by using a differentiable dMRI simulator to iteratively update the mesh.

Please refer to our arxiv for more details.

Here is an overview of iterative reconstruction of a bent axon by ReMiDi at different gradient descent iterations. The ground-truth mesh is shown to the left. Top row: Evolution of the triangulated mesh surface. Bottom row: Corresponding point cloud representations of mesh vertices. The color scale shows Chamfer distance (lower values are better).

Method

Above is the ReMiDi-based dMRI reconstruction pipeline. A 3D mesh decoded by a Spectral Auto-Encoder (SAE) is given as input to the differentiable dMRI simulator. The final loss function between the simulated dMRI signal and a reference signal is backpropagated to the latent of the SAE, and the mesh is iteratively updated with gradient-based learning.

Running Experiments

Current setup provides three different reconstruction experiments: bending, beading, and fanning. Each experiment can be run using its corresponding configuration file.

Using Docker

The recommended way to run ReMiDi is using our Docker image:

docker pull prattk/remidi:v1.0.1

Reconstruction Examples

1. Bending Reconstruction:

python3 bending_reconstruction.py config/bending_reconstruction_conf.json

2. Beading Reconstruction:

python3 beading_reconstruction.py config/beading_reconstruction_conf.json

3. Fanning Reconstruction:

python3 fanning_reconstruction.py config/fanning_reconstruction_conf.json

Output

The reconstruction process generates:

• Intermediate mesh files in .html format for visualization
• Image files showing the reconstruction progress
• Video files for the entire reconstruction process will also be generated and saved in the same folder.

These outputs are automatically saved in the output directory specified in the configuration files.

Citation

If you use ReMiDi in your research, please cite our paper:

@misc{khole2025remidireconstructionmicrostructureusing,
      title={ReMiDi: Reconstruction of Microstructure Using a Differentiable Diffusion MRI Simulator}, 
      author={Prathamesh Pradeep Khole and Zahra Kais Petiwala and Shri Prathaa Magesh and Ehsan Mirafzali and Utkarsh Gupta and Jing-Rebecca Li and Andrada Ianus and Razvan Marinescu},
      year={2025},
      eprint={2502.01988},
      archivePrefix={arXiv},
      primaryClass={eess.IV},
      url={https://arxiv.org/abs/2502.01988}, 
}