Source code for ClearMap.Scripts.tube_map_new_api
"""
tube_map_new_api
================
Headless entry point for the TubeMap pipeline (vasculature binarization,
graph construction, and annotation).
This script replaces the deprecated :mod:`ClearMap.Scripts.TubeMap`.
It initialises a sample from the experiment directory, runs stitching and
atlas registration, then binarizes each vessel channel, merges the binary
masks, and builds an annotated vasculature graph.
Usage
-----
.. code-block:: bash
conda activate ClearMap3.1
python -m ClearMap.Scripts.tube_map_new_api /path/to/experiment
The experiment directory must contain a ``sample.yml`` config file (created
by the GUI or by copying from ``~/.clearmap/defaults/``). All pipeline
parameters are read from the YAML files in that directory; no editing of
this script is required.
Steps performed
---------------
1. **Stitching** — assembles tiles into a single volume for each channel.
2. **Registration** — resamples the autofluorescence channel and aligns it
to the atlas via Elastix.
3. **Binarization** (per TubeMap channel) —
a. :meth:`~ClearMap.pipeline_orchestrators.tube_map.BinaryVesselProcessor.binarize_channel`
— multi-path thresholding.
b. :meth:`~ClearMap.pipeline_orchestrators.tube_map.BinaryVesselProcessor.smooth_channel`
— topology-preserving binary smoothing.
c. :meth:`~ClearMap.pipeline_orchestrators.tube_map.BinaryVesselProcessor.fill_channel`
— parallel 3-D binary hole filling.
d. :meth:`~ClearMap.pipeline_orchestrators.tube_map.BinaryVesselProcessor.deep_fill_channel`
— CNN-based hollow-tube filling (requires GPU with ≥ 24 GB VRAM).
4. :meth:`~ClearMap.pipeline_orchestrators.tube_map.BinaryVesselProcessor.combine_binary`
— logical-OR merge of all channel masks into a single combined binary.
5. **Graph construction** —
a. :meth:`~ClearMap.pipeline_orchestrators.tube_map.VesselGraphProcessor.pre_process`
— skeletonize → build raw graph → clean → reduce → register to atlas.
b. :meth:`~ClearMap.pipeline_orchestrators.tube_map.VesselGraphProcessor.post_process`
— iterative artery/vein tracing and capillary removal (if artery channel present).
c. :meth:`~ClearMap.pipeline_orchestrators.tube_map.VesselGraphProcessor.voxelize`
— rasterise graph vertices into a branch density volume.
Outputs (written to the experiment directory via the workspace)
---------------------------------------------------------------
* ``binary.npy`` / ``binary_smoothed.npy`` / … — intermediate binary masks
* ``binary_combined.npy`` — merged vessel mask
* ``skeleton.npy`` — skeletonized binary
* ``graph_raw.gt`` / ``graph_cleaned.gt`` / ``graph_reduced.gt`` / ``graph_annotated.gt``
— graph at each construction stage
* ``density_branches.tif`` — voxelized vessel density in atlas space
* ``vertices.feather`` — vertex table with coordinates, radii, and atlas labels
See also
--------
:class:`~ClearMap.pipeline_orchestrators.tube_map.BinaryVesselProcessor` :
Binarization worker.
:class:`~ClearMap.pipeline_orchestrators.tube_map.VesselGraphProcessor` :
Graph construction and annotation worker.
:func:`~ClearMap.pipeline_orchestrators.utils.init_sample_manager_and_processors` :
Convenience factory used to initialise all standard workers.
:doc:`/tubemap` :
Full TubeMap pipeline documentation.
"""
import sys
from ClearMap.pipeline_orchestrators.utils import init_sample_manager_and_processors
from ClearMap.pipeline_orchestrators.tube_map import BinaryVesselProcessor, VesselGraphProcessor
from ClearMap.Scripts.align_new_api import stitch, register, plot_registration_results
[docs]
def main(src_directory):
orchestrators = init_sample_manager_and_processors(src_directory)
sample_manager = orchestrators['sample_manager']
stitcher = orchestrators['stitcher']
registration_processor = orchestrators['registration_processor']
stitch(stitcher)
stitcher.plot_stitching_results(mode='overlay')
register(registration_processor)
plot_registration_results(registration_processor, sample_manager.alignment_reference_channel)
binary_vessel_processor = BinaryVesselProcessor(sample_manager,
config_coordinator=sample_manager.cfg_coordinator)
for channel in sample_manager.get_channels_by_pipeline('TubeMap', as_list=True):
binary_vessel_processor.binarize_channel(channel)
binary_vessel_processor.smooth_channel(channel)
binary_vessel_processor.fill_channel(channel)
binary_vessel_processor.deep_fill_channel(channel)
binary_vessel_processor.combine_binary()
# binary_vessel_processor.plot_combined(arrange=True)
vessel_graph_processor = VesselGraphProcessor(sample_manager, config_coordinator=sample_manager.cfg_coordinator,
registration_processor=registration_processor)
vessel_graph_processor.pre_process()
# TODO: slice
vessel_graph_processor.post_process()
vessel_graph_processor.voxelize()
# vessel_graph_processor.plot_voxelization(None)
if __name__ == '__main__':
main(sys.argv[1])