Training-Free Condition Video Diffusion Models for single frame Spatial-Semantic Echocardiogram Synthesis
Official implementation of the paper "Training-Free Condition Video Diffusion Models for single frame Spatial-Semantic Echocardiogram Synthesis" (MICCAI 2024).
- Python 3.6+
- PyTorch 1.7.1+
- CUDA 10.1+
git clone
cd echocardiography-video-synthesis
pip install -r requirements.txtCAMUS dataset can be downloaded from here.
You have to extract the dataset in the data folder. which have the following structure:
camus
├── images
│ ├── patient0001
│ │ ├── 0000.png
│ │ ├── 0001.png
│ │ ├── ...
│ │ └── 0020.png
│ ├── patient0002
│ │ ├── 0000.png
│ │ ├── 0001.png
│ │ ├── ...
│ │ └── 0020.png
│ ├── ...
│ └── patient0020
│ ├── 0000.png
│ ├── 0001.png
│ ├── ...
│ └── 0020.png
├── seg_maps_cone
│ ├── patient0001
│ │ ├── 0000.png
│ │ ├── 0001.png
│ │ ├── ...
│ │ └── 0020.png
│ ├── patient0002
│ │ ├── 0000.png
│ │ ├── 0001.png
│ │ ├── ...
│ │ └── 0020.png
│ ├── ...
│ └── patient0020
│ ├── 0000.png
│ ├── 0001.png
│ ├── ...
│ └── 0020.png
├── train.txt
├── val.txt
└── test.txt
# Training
```bash
python train.py --data_dir data/camus
python sample.py --data_dir data/camus --output_dir samples --is_sdedit Falseit glob some of example segmentation maps from visualize folder and generate the output images in the visualize folder as well.
If you find this code useful for your research, please cite our paper:
@inproceedings{phi2024trainingfree,
title={Training-Free Condition Video Diffusion Models for single frame Spatial-Semantic Echocardiogram Synthesis},
author={Van Phi, Nguyen and Tri Nhan, Luong Ha and Huy Hieu, Pham and Quoc Long, Tran},
booktitle={MICCAI 2024 Medical Image Computing and Computer Assisted Intervention},
year={2024}
}
This code is based on the Video Diffusion Model