pakheiyeung/PlaneInVol

A network, trained with just registered volumes, to predict the 3D location of 2D freehand ultrasound fetal brain images and video

Python

Learning to Map 2D Ultrasound Images into 3D Space with Minimal Human Annotation

This repository contains the codes (in PyTorch) for the framework introduced in the following paper:

Learning to Map 2D Ultrasound Images into 3D Space with Minimal Human Annotation [Paper] [Project Page]

@article{yeung2021learning,
	title = {Learning to map 2D ultrasound images into 3D space with minimal human annotation},
	author = {Yeung, Pak-Hei and Aliasi, Moska and Papageorghiou, Aris T and Haak, Monique and Xie, Weidi and Namburete, Ana IL},
	journal = {Medical Image Analysis},
	volume = {70},
	pages = {101998},
	year = {2021},
	publisher = {Elsevier}
}

Contents

Dependencies
Train
Inference on Video Data

Dependencies

Python (3.6), other versions should also work
PyTorch (1.6), other versions should also work
scipy
cv2
skimage
imgaug

Train

Modify the train_github.py for importing your dataset according to the instructions in the file.
Use the following command to train:
```
python train_github.py
```

Inference on Video Data

Post (predict after the whole video is acquired)

Modify the inference_github.py according to the instructions in the file.
Use the following command to train:
```
python inference_github.py
```

Real-time

To do, but it would be an easy modification based on the inference_github.py.