WorldWarp MRI Reconstruction

# WorldWarp MRI Reconstruction ## Description This project aims to adapt the WorldWarp asynchronous video diffusion method to enhance 3D reconstruction of irregularly sampled MRI sequences. By leveraging the method's ability to handle occlusions and irregular sampling, we hope to improve the fidelity of reconstructed MRI images. ## Research Hypothesis Applying WorldWarp's asynchronous video diffusion to medical imaging can enhance the 3D reconstruction of irregularly sampled MRI sequences. ## Implementation Approach - Adapt the WorldWarp architecture for 3D MRI data. - Use datasets like fastMRI and IXI for training. - Evaluate the model using SSIM and RMSE metrics. ## Setup Instructions 1. Clone the repository: `git clone ` 2. Navigate to the project directory: `cd WorldWarp_MRI_Reconstruction` 3. Install dependencies: `pip install -r requirements.txt` ## Usage Examples - Run training: `python src/train.py` - Run evaluation: `python src/evaluate.py` ## Expected Results Improved SSIM and reduced RMSE in reconstructed MRI volumes compared to standard methods. ## References - WorldWarp: Propagating 3D Geometry with Asynchronous Video Diffusion [Paper](http://arxiv.org/abs/2512.19678v1)