Skip to content

Commit

Permalink
fix dim order for workflow widget saving
Browse files Browse the repository at this point in the history 
fix dim order saving in workflow widget to accomdote XY input only. Was previously parsing as TZCYX, which was problem because OmeTiffWriter wants TCZYX and had trouble with channel name metadata
  • Loading branch information
@TimMonko
TimMonko committed Aug 25, 2023
1 parent 5f434ae commit 7c9aa27
Showing 1 changed file with 50 additions and 25 deletions.
75 changes: 50 additions & 25 deletions src/napari_ndev/_widget.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -78,22 +78,27 @@ def _image_info():
result_directory=dict(widget_type="FileEdit", mode="d"),
channel_list=dict(widget_type="Select", choices=[]),
keep_scenes=dict(widget_type="LineEdit"),
project_combo=dict(widget_type="ComboBox", choices=["No Projection","np.max","np.sum"]),
project_combo=dict(
widget_type="ComboBox", choices=["No Projection", "np.max", "np.sum"]
),
)
def batch_utilities(
image_directory=pathlib.Path(),
result_directory=pathlib.Path(),
channel_list: str = [],
keep_scenes: str = "",
#project_bool: bool = False,
# project_bool: bool = False,
project_combo: str = "No Projection",
X_range: slice = slice(0, 1, 1),
Y_range: slice = slice(0, 1, 1),
Z_range: slice = slice(0, 1, 1),
):
"""Batch Utilities
Quick adjustments to apply to a batch of images and save the resulting images in an output folder. Intended for adjustments either too simple (e.g. max projection, saving only certain channels) or not possible (e.g. cropping) with napari-workflows / batch-workflow widgets.
Quick adjustments to apply to a batch of images and save the resulting
images in an output folder. Intended for adjustments either too simple
(e.g. max projection, saving only certain channels) or not possible
(e.g. cropping) with napari-workflows / batch-workflow widgets.
Parameters
----------
Expand Down Expand Up @@ -167,6 +172,7 @@ def batch_utilities(
# save the image
save_name = str(file + "ome.tif")
save_uri = result_directory / save_name

OmeTiffWriter.save(
data=result_stack,
uri=save_uri,
Expand Down Expand Up @@ -201,7 +207,7 @@ def annotation_saver(
selecting of the intended label layer and image layer, as well as
the prefix for the output folders. These output folders can already
exist: mkdir(exist_ok=True), so should be used to save multiple
images within the same folder group. Images are saved as ome.tif
images within the same folder group. Images are saved as ome.tif
with aicsimageio.OmeTiffWriter
Parameters
Expand All @@ -215,8 +221,8 @@ def annotation_saver(
output_folder_prefix : str
Prefix to _images or _labels folder created in `file_directory`,
by default "Annotated"
save_suffix : str
File ending can be changed if needed for interoperability, but still
save_suffix : str
File ending can be changed if needed for interoperability, but still
saves as an OME-TIFF with aicsimageio.OmeTiffWriter,
by default "ome.tif"
Expand Down Expand Up @@ -322,12 +328,12 @@ def batch_workflow(
):
"""Batch Workflow widget using napari-workflow metadata file
Load a napari-workflow metadata file and show the original roots.
Load a napari-workflow metadata file and show the original roots.
Select an image directory to populate the root dropdowns with
channels from the first image read by aicsimageio. Then, select these
channels in the dropdown to match the Roots in the proper order.
If there are less roots than displayed, leave dropdown as
'-----' which represents python None.
channels from the first image read by aicsimageio. Then, select these
channels in the dropdown to match the Roots in the proper order.
If there are less roots than displayed, leave dropdown as
'-----' which represents python None.
Parameters
----------
Expand All @@ -342,14 +348,17 @@ def batch_workflow(
root_0, root_1, root_2, root_3, root_4 : str, optional
_description_, by default None
img_dims : str, optional
Can be changed, if necessary, to account for different image shapes by default None
Can be changed, if necessary, to account for different image
shapes by default None
keep_original_images : bool, optional
Stack original images with result images prior to saving, by default True
Stack original images with result images prior to saving,
by default True
Returns
-------
None
Resulting image stacks are saved to result_directory after workflow processing
Resulting image stacks are saved to result_directory after workflow
processing
"""
image_list = os.listdir(image_directory)

Expand Down Expand Up @@ -398,10 +407,17 @@ def batch_workflow(
save_name = str(file + ".ome.tif")
save_uri = result_directory / save_name

# Need to explicitly order CYX if dims are just XY else the stack
# will be in the order of TZCYX but the saver won't know
if img_dims == "YX":
save_dim_order = "CYX"
else:
save_dim_order = None

OmeTiffWriter.save(
data=result_stack,
uri=save_uri,
# dim_order=img.dims.order,
dim_order=save_dim_order, # this was previously commented out
channel_names=result_names,
physical_pixel_sizes=img.physical_pixel_sizes,
)
Expand Down Expand Up @@ -464,21 +480,25 @@ def batch_training(
cl_directory : pathlib.Path
Location to save apoc classifier ".cl" file
cl_filename : str, optional
Filename to save apoc classifier, end with ".cl", by default "classifier.cl"
Filename to save apoc classifier, end with ".cl", by default
"classifier.cl"
cl_type : str, optional
Choose between Pixel or Object Classifier, by default Pixel Classifier
cl_forests : int, optional
Number of random forests, by default 2
cl_trees : int, optional
Number of trees, by default 100
predefined_features : str, optional
Allows selection of `apoc.PredefinedFeatureSets`, by default custom, requires input of `custom_features`
Allows selection of `apoc.PredefinedFeatureSets`, by default custom,
requires input of `custom_features`
custom_features : str, optional
Space-separated string of pyclesperanto filters, by default None
channel_list : list
Select channels to pass into the classifier to serve as bases for features, by default []
Select channels to pass into the classifier to serve as bases for
features, by default []
cl_label_id : int, optional
Label id number if using Object Classifier for `cl_type`, by default 2
Label id number if using Object Classifier for `cl_type`,
by default 2
img_dims : str, optional
Image dimensions read from aicsimageio metadata, by default None
Expand All @@ -489,9 +509,10 @@ def batch_training(
Notes
-----
Accelerated pixel object classifier information from:
Accelerated pixel object classifier information from:
https://github.com/haesleinhuepf/apoc
Predefined features from https://github.com/haesleinhuepf/apoc/blob/main/demo/feature_stacks.ipynb
Predefined features from
https://github.com/haesleinhuepf/apoc/blob/main/demo/feature_stacks.ipynb
"""
image_list = os.listdir(image_directory)
label_list = os.listdir(label_directory)
Expand All @@ -515,8 +536,11 @@ def batch_training(
num_ensembles=cl_trees,
)

# use an enumerate so that the index of the file can be used to extract the proper label file, in case the image and label files do not have matching names
# This could be problematic if the images aren't sorted the same way, but should generally be ok
# use an enumerate so that the index of the file can be used to extract
# the proper label file, in case the image and label files do not have
# matching names
# This could be problematic if the images aren't sorted the same way,
# but should generally be ok
for idx, file in enumerate(tqdm(image_list, label="progress")):

image_stack = []
Expand Down Expand Up @@ -585,7 +609,7 @@ def batch_predict(
channel_list: str = [],
img_dims: str = None,
):
"""Train APOC (Accelerated-Pixel-Object-Classifiers) on a folder of
"""Predict APOC (Accelerated-Pixel-Object-Classifiers) on a folder of
images and labels.
Parameters
Expand All @@ -599,7 +623,8 @@ def batch_predict(
cl_type : str, optional
Choose between Pixel or Object Classifier, by default Pixel Classifier
channel_list : list
Select channels to pass into the classifier to serve as bases for features, by default []
Select channels to pass into the classifier to serve as bases for
features, by default []
cl_label_id : int, optional
Label id number if using Object Classifier for `cl_type`, by default 2
img_dims : str, optional
Expand Down

0 comments on commit 7c9aa27

Please sign in to comment.