FUNSR: Neural Implicit Surface Reconstruction of Freehand 3D Ultrasound Volume with Geometric Constraint

Paper | Arxiv

The official implementation code for Medical Image Analysis paper: Neural Implicit Surface Reconstruction of Freehand 3D Ultrasound Volume with Geometric Constraint

FUNSR is a Freehand 3D Ultrasound Neural Surface Reconstruction method built on neural implicit signed distance functions for freehand 3D ultrasound imaging and surface rendering.

Demo

• Example results of publicly available Dataset #4.

Usage

Our code is implemented in Python 3.8, PyTorch 1.12.1 and CUDA 11.6

Install Dependencies

conda create -n funsr python=3.8
conda activate funsr
conda install pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cudatoolkit=11.6 -c pytorch -c conda-forge
pip install tqdm pyhocon==0.3.57 trimesh PyMCubes scipy
pip install matplotlib

Data Preparation

• Put the point cloud data on ./data.

• The point cloud data format is in *.ply.

Run FUNSR

python run_normalizedSpace.py --gpu 0 --conf confs/conf.conf --dataname case000070.nii_ds  --dir case000070.nii_ds

Contact

For any queries, please contact chenhb[at]shanghaitech.edu.cn.

References

The reference codes are from the following links. We appreciate all the contributors.

• NeuralPull: https://github.com/mabaorui/NeuralPull-Pytorch

• GenSDF: https://github.com/princeton-computational-imaging/gensdf

• PU-GAN: https://github.com/liruihui/PU-GAN