Reuse masks and saved meshes
============================

Resume at different points in the workflow, provided the inputs and saved
geometry belong to the same fields and calibration.

Reuse existing masks
--------------------

.. code-block:: python

   from bactoscoop import ImageCollection
   ic = ImageCollection(r"C:\path\to\data", px=0.065)
   ic.create_image_objects(phase_channel="C1")
   ic.batch_process_mesh(phase_channel="C1", save_data=True)

``create_image_objects`` loads missing phase images and masks, validates their
field identities, and creates new image objects. In the tutorials, set
``RUN_SEGMENTATION=False`` to take this route.

Reuse geometry
--------------

.. code-block:: python

   ic = ImageCollection(r"C:\path\to\data", px=0.065)
   ic.batch_load_mesh("data_curated_meshdata.pkl", phase_channel="C1")
   # Continue with channel detection and feature extraction.

``batch_load_mesh`` accepts pickle and reloadable mesh Parquet. ``pkl_path`` can
specify the directory containing the mesh file; its default is the image folder.
The image/mask files are still needed to reconstruct image objects and measure
signals. The file needs compatible geometry columns and field identities.

State after loading
-------------------

Reloading creates a new image/cell population and clears previous feature and
merged tables. Saved mesh tables preserve geometry, not all channel caches,
detected objects, or feature dictionaries. Redetect objects and calculate the
requested features after reloading.

Verify ``px`` against the original experiment rather than assuming the saved
geometry file carries all acquisition settings. See
:py:meth:`bactoscoop.imagecollection.ImageCollection.batch_load_mesh`.