Task-based Adaptive Transmit Beamforming for Efficient Ultrasound Quantification

The repo contains the code for the paper Task-based Adaptive Transmit Beamforming for Efficient Ultrasound Quantification.

Online transmit-scheme optimization designed to minimize uncertainty about downstream measurement parameters, enabling high quality estimates using a small fraction of the transmit events typically necessary.

Find the weights of our models on Huggingface:

• EchoNetLVH Segmentation Model
• EchoNetLVH Diffusion Model

Setup code

Install zea, the cognitive ultrasound toolbox.

pip install "zea==0.0.4"

or use the submodule in this repo:

git submodule update --init --recursive
pip install -e zea

Install other dependencies for this repo:

KERAS_VER=$(python3 -c "import keras; print(keras.__version__)")
pip install tf2jax==0.3.6 pandas jaxwt dm-pix jax
pip install keras==${KERAS_VER}
cp .env.example .env
touch users.yaml # edit!

Dataset

• Download the EchoNetLVH dataset.

• Convert the dataset to the polar format using the zea conversion scripts described here.

Example Notebook

See the zea example notebook implementing the task-based perception action loop, here.