Overview of the zea data processing workflow.
① Generate ``zea.Parameters`` from a zea file using ``File.load_parameters()`` — this combines
scan and probe information from the file to compute all parameters needed for beamforming.
② Optionally apply additional parameter overrides from a ``config.yaml``.
③ Initialise the pipeline, either from a ``config.yaml`` or manually in code.
④ Pass data and parameters to the pipeline for processing.
⑤ Visualise your outputs.
.. code-block:: python
import zea
# setting up cpu / gpu usage
zea.init_device()
# loading a config file from Hugging Face, but can also load a local config file
config = zea.Config.from_path("hf://zeahub/configs/config_picmus_rf.yaml")
path = config.data.dataset_folder + "/" + config.data.file_path
with zea.File(path) as file:
data = file.data.raw_data[0]
# load the merged probe + scan parameters and apply config overrides
parameters = file.load_parameters()
parameters.update(**config.parameters)
# using the pipeline as specified in the config file
pipeline = zea.Pipeline.from_config(
config,
with_batch_dim=False,
)
# prepare the inputs (converts the needed parameters to tensors)
inputs = pipeline.prepare_parameters(parameters)
# running the pipeline!
image = pipeline(data=data, **inputs)["data"]
# show the image
image = zea.display.to_8bit(image)
image.show()
Similarly, we can easily load one of the pretrained models from the :mod:`zea.models` module and use it for inference.
.. code-block:: python
import keras
import zea
from zea.models.lv_segmentation import AugmentedCamusSeg
zea.init_device()
# NOTE: for this model you need: `pip install onnxruntime`
model = AugmentedCamusSeg.from_preset("augmented_camus_seg")
with zea.Dataset("hf://zeahub/camus-sample/") as dataset:
file = dataset[0]
image = file.data.image_sc[0]
# Resize to 256x256 and normalize to [-1, 1] as expected by the model
image = keras.ops.image.resize(image[None, ..., None], (256, 256))[0, ..., 0]
image = zea.func.translate(image, range_to=[-1, 1])
# Run model: expects NCHW format (batch, channels, height, width)
image = keras.ops.convert_to_numpy(image)
predictions = model(image[None, None]) # (1, N, H, W)
# Compute class assignments via argmax to get mutually exclusive masks
class_map = predictions[0].argmax(axis=0) # (H, W) — each pixel assigned to one class
# Derive specific masks: class 1 = LV, class 2 = myocardium
lv_mask = class_map == 1
myo_mask = class_map == 2
image = zea.display.to_8bit(image, dynamic_range=(-1, 1))
masks = [
zea.display.to_8bit(lv_mask, dynamic_range=(0, 1)),
zea.display.to_8bit(myo_mask, dynamic_range=(0, 1)),
]
result = zea.display.overlay_masks(image, masks, alpha=0.5)
result.show()
``zea`` also provides a simple command line interface (CLI) to quickly visualize a ``zea`` data file.
.. code-block:: shell
zea --config configs/config_picmus_rf.yaml
Installation
------------
A simple pip command will install the latest version of ``zea`` from `PyPI