Assessing Test-time Variability for Interactive 3D Medical Image Segmentation with Diverse Point Prompts
Introduction
Our code can be seamlessly integrated into any SAM-based or interactive methods for medical image segmentation. We assessed the test-time variability using the pretrained ProMISe, where the models can be found here.
Installation
conda create -n promise python=3.9
conda activate promise
(Optional): sudo install git
pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 --extra-index-url https://download.pytorch.org/whl/cu113 # install pytorch
pip install git+https://github.com/facebookresearch/segment-anything.git # install segment anything packages
pip install git+https://github.com/deepmind/surface-distance.git # for normalized surface dice (NSD) evaluation
pip install -r requirements.txt
Datasets
Here are the datasets that we used in our experiments, which are modified based on the original public datasets from Medical Segmentation Decathlon. We task 10 for colon tumor segmentations.
Use
python ./boundary_selection/boundary_selection.py
An example case is provided under ./boundary_selection, please see the example.png
contact
Please shoot an email to hao.li.1@vanderbilt.edu for any questions, and I am always happy to help! :).
@article{li2023assessing,
title={Assessing Test-time Variability for Interactive 3D Medical Image Segmentation with Diverse Point Prompts},
author={Li, Hao and Liu, Han and Hu, Dewei and Wang, Jiacheng and Oguz, Ipek},
journal={arXiv preprint arXiv:2311.07806},
year={2023}
}
@article{li2023promise,
title={Promise: Prompt-driven 3D Medical Image Segmentation Using Pretrained Image Foundation Models},
author={Li, Hao and Liu, Han and Hu, Dewei and Wang, Jiacheng and Oguz, Ipek},
journal={arXiv preprint arXiv:2310.19721},
year={2023}
}