LiverUSRecon: Automatic 3D Reconstruction and Volumetry of the Liver with a Few Partial Ultrasound Scans
This website holds information for LiverUSRecon: Automatic 3D Reconstruction and Volumetry of the Liver with a Few Partial Ultrasound Scans
- Codes are available at LiverUSRecon.
- If you need the data, please email to Kaushalya Sivayogaraj.
- Weights are available at Segmentation model and Reconstruction model.
Kaushalya Sivayogaraj, Sahan T. Guruge, Udari Liyanage, Jeevani Udupihille, Saroj Jayasinghe, Gerard Fernando, Ranga Rodrigo, Rukshani Liyanaarachchi
3D reconstruction of the liver for volume measurement and 3D visual shape analysis using an accessible medical imaging modality like ultrasound (US) imaging is important. We present the first method capable of reconstructing liver from few partial Ultrasound scans aquired at midline, midclavicular line and anterior-auxillay line. To the best of our knowledge, this is the first automated deep learning method that calculates the liver volume from three incomplete 2D US scans. Further, we introduce a new US liver database with parallel, annotated CT scans comprising 134 scans.Our volumetry results are statistically closer to the ground-truth volumes obtained from CT scans than the volumes computed by radiologists using the Childs’ method.
- Download R50-ViT-B_16 models in this link: R50-ViT-B_16
- Move the downloaded model to folder
./model/vit_checkpoint/imagenet21k/
and rename it toR50-ViT-B_16.npz
- Download the pretrained segmentation and reconstruction models from pretrained segmentation models pretrained reconstruction models and move it to folder named "models" under the results folder
- Please email to Kaushalya Sivayogaraj to collect the inference datasets.
- Download SSM information shape parameters, mean shape, pca ratio and normalization info
- Once you download the SSM information, place it in the folder
./SSM/
- Create an environment with python=3.7 and install the dependencies.
pip install -r requirements.txt
- Run the inference_liverusrecon script on the downloaded dataset.
CUDA_VISIBLE_DEVICES=0 python inference_liverusrecon.py --inference {dataset path} --save {results path} --ssm_info {ssm_info path}
Code is covered under the GNU Affero General Public License version 3.0
You should have received a copy of the GNU Affero General Public License along with the code. If not, see https://www.gnu.org/licenses/.
ML Weights are licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.
You should have received a copy of the license along with this work. If not, see https://creativecommons.org/licenses/by-nc-nd/3.0/.
Patient data is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.
You should have received a copy of the license along with this work. If not, see https://creativecommons.org/licenses/by-nc-nd/3.0/.
If you find this project or this repository useful, please consider cite:
@misc{sivayogaraj2024liverusreconautomatic3dreconstruction,
title={LiverUSRecon: Automatic 3D Reconstruction and Volumetry of the Liver with a Few Partial Ultrasound Scans},
author={Kaushalya Sivayogaraj and Sahan T. Guruge and Udari Liyanage and Jeevani Udupihille and Saroj Jayasinghe and Gerard Fernando and Ranga Rodrigo and M. Rukshani Liyanaarachchi},
year={2024},
eprint={2406.19336},
archivePrefix={arXiv},
primaryClass={eess.IV},
url={https://arxiv.org/abs/2406.19336},
}