PHOCUS: Physics-Based Deconvolution for Ultrasound Resolution Enhancement

Pre-Print

Felix Duelmer^1,2,3, Walter Simson⁴, Mohammad Farid Azampour^1,2, Magdalena Wysocki^1,2, Angelos Karlas^1,5, Nassir Navab^1,2,

¹Technical University of Munich, Munich, Germany, ²Munich Centre for Machine Learning, Munich, Germany, ³Helmholtz Centre Munich, Neuherberg, Germany, ⁴Stanford University, Stanford, USA, ⁵German Centre for Cardiovascular Research, Munich, Germany
Stanford University

This is the official implementation of the paper "PHOCUS: Physics-Based Deconvolution for Ultrasound Resolution Enhancement".

Pipeline

Getting Started

You can install the required Python dependencies like this:

conda env create -f env.yml
conda activate phocus

Reproducing results

The hyperparameters set in the config files can be used to reproduce the results from the paper. The data from the paper can be found under data/

Run the program with this command:

python3 run_nerf.py --config configs/conf_cirs_synthetic_wires.txt

Citation

Will be updated soon.

@inproceedings{duelmer2024phocus,
  title={PHOCUS: Physics-Based Deconvolution for Ultrasound Resolution Enhancement},
  author={Duelmer, Felix and Simson, Walter and Azampour, Mohammad Farid and Wysocki, Magdalena and Karlas, Angelos and Navab, Nassir},
  booktitle={Annual Conference on Medical Image Understanding and Analysis},
  organization={Springer}
}

Acknowledgments

Large parts of the code are from the HashNeRF-pytorch implementation. See ACKNOWLEDGEMENTS.