DopUS-Net: Quality-Aware Robotic Ultrasound Imaging Based on Doppler Signal

Pytorch implementation of segmentation based on Ultrasound B-Mode and Doppler images. Accepted at IEEE Transactions on Automation Science and Engineering.

DopUS-Net: Quality-Aware Robotic Ultrasound Imaging Based on Doppler Signal
Zhongliang Jiang*, Felix Dülmer*, Nassir Navab
Technical University of Munich (TUM)

*denotes equal contribution

Installation

pip install -r requirements.txt

Training

To run the latest version of the DopUs-Network please run:

• train_segmentation.py -c configs/config_dopus_v4.json

Visualization

To visualize the training process please launch a local visdom server:

• python -m visdom.server