Creation (pymia.data.creation package)

Callback (pymia.data.creation.callback module)

class pymia.data.creation.callback.Callback

Bases: object

Base class for the interaction with the dataset creation.

Implementations of the Callback class can be provided to Traverser.traverse() in order to write/process specific information of the original data.

on_end(params: dict)

Called at the end of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser. The provided parameters will differ from Callback.on_subject().

on_start(params: dict)

Called at the beginning of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser. The provided parameters will differ from Callback.on_subject().

on_subject(params: dict)

Called for each subject of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser containing subject specific information and data.

class pymia.data.creation.callback.ComposeCallback(callbacks: List[Callback])

Bases: Callback

Composes many Callback instances and behaves like an single Callback instance.

This class allows passing multiple Callback to Traverser.traverse().

Parameters:

callbacks (list) – A list of Callback instances.

on_end(params: dict)

Called at the end of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser. The provided parameters will differ from Callback.on_subject().

on_start(params: dict)

Called at the beginning of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser. The provided parameters will differ from Callback.on_subject().

on_subject(params: dict)

Called for each subject of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser containing subject specific information and data.

class pymia.data.creation.callback.MonitoringCallback

Bases: Callback

Callback that monitors the dataset creation process by logging the progress to the console.

on_end(params: dict)

Called at the end of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser. The provided parameters will differ from Callback.on_subject().

on_start(params: dict)

Called at the beginning of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser. The provided parameters will differ from Callback.on_subject().

on_subject(params: dict)

Called for each subject of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser containing subject specific information and data.

class pymia.data.creation.callback.WriteDataCallback(writer: Writer)

Bases: Callback

Callback that writes the raw data to the dataset.

Parameters:

writer (.creation.writer.Writer) – The writer used to write the data.

on_subject(params: dict)

Called for each subject of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser containing subject specific information and data.

class pymia.data.creation.callback.WriteEssentialCallback(writer: Writer)

Bases: Callback

Callback that writes the essential information to the dataset.

Parameters:

writer (.creation.writer.Writer) – The writer used to write the data.

on_start(params: dict)

Called at the beginning of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser. The provided parameters will differ from Callback.on_subject().

on_subject(params: dict)

Called for each subject of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser containing subject specific information and data.

class pymia.data.creation.callback.WriteFilesCallback(writer: Writer)

Bases: Callback

Callback that writes the file names to the dataset.

Parameters:

writer (.creation.writer.Writer) – The writer used to write the data.

on_start(params: dict)

Called at the beginning of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser. The provided parameters will differ from Callback.on_subject().

on_subject(params: dict)

Called for each subject of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser containing subject specific information and data.

class pymia.data.creation.callback.WriteImageInformationCallback(writer: Writer, category='images')

Bases: Callback

Callback that writes the image information (shape, origin, direction, spacing) to the dataset.

Parameters:

• writer (.creation.writer.Writer) – The writer used to write the data.

• category (str) – The category from which to extract the information from.

on_start(params: dict)

Called at the beginning of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser. The provided parameters will differ from Callback.on_subject().

on_subject(params: dict)

Called for each subject of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser containing subject specific information and data.

class pymia.data.creation.callback.WriteNamesCallback(writer: Writer)

Bases: Callback

Callback that writes the names of the category entries to the dataset.

Parameters:

writer (.creation.writer.Writer) – The writer used to write the data.

on_start(params: dict)

Called at the beginning of Traverser.traverse().

Parameters:

params (dict) – Parameters provided by the Traverser. The provided parameters will differ from Callback.on_subject().

pymia.data.creation.callback.get_default_callbacks(writer: Writer, meta_only=False) → ComposeCallback

Provides a selection of commonly used callbacks to write the most important information to the dataset.

Parameters:

• writer (.creation.writer.Writer) – The writer used to write the data.

• meta_only (bool) – Whether only callbacks for a metadata dataset creation should be returned.

Returns:

The composed selection of common callbacks.

Return type:

Callback

File loader (pymia.data.creation.fileloader module)

class pymia.data.creation.fileloader.Load

Bases: ABC

Interface for loading the data during the dataset creation in Traverser.traverse()

abstract __call__(file_name: str, id_: str, category: str, subject_id: str) → Tuple[ndarray, ImageProperties | None]

Loads the data from the file system according to the implementation.

Parameters:

• file_name (str) – Path to the corresponding data.

• id (str) – Identifier for the entry of the category, e.g., “Flair”.

• category (str) – Name of the category, e.g., ‘images’.

• subject_id (str) – Identifier of the current subject.

Returns:

A numpy array containing the loaded data and ImageProperties describing the data. ImageProperties is None if the loaded data does not contain further properties.

Return type:

tuple

class pymia.data.creation.fileloader.LoadDefault

Bases: Load

The default loader.

It loads every data item (id/entry, category) for each subject as sitk.Image and the corresponding ImageProperties.

Traverser (pymia.data.creation.traverser module)

class pymia.data.creation.traverser.Traverser(categories: str | Tuple[str, ...] | None = None)

Bases: object

Class managing the dataset creation process.

Parameters:

categories (str or tuple of str) – The categories to traverse. If None, then all categories of a SubjectFile will be traversed.

traverse(subject_files: ~typing.List[~pymia.data.subjectfile.SubjectFile], load=<pymia.data.creation.fileloader.LoadDefault object>, callback: ~pymia.data.creation.callback.Callback | None = None, transform: ~pymia.data.transformation.Transform | None = None, concat_fn=<function default_concat>)

Controls the actual dataset creation. It goes through the file list, loads the files, applies transformation to the data, and calls the callbacks to do the storing (or other stuff).

Parameters:

• subject_files (list) – list of SubjectFile to be processes.

• load (callable) – A load function or Load instance that performs the data loading

• callback (.Callback) – A callback or composed (ComposeCallback) callback performing the storage of the loaded data (and other things such as logging).

• transform (.Transform) – Transformation to be applied to the data after loading and before Callback.on_subject() is called

• concat_fn (callable) – Function that concatenates all the entries of a category (e.g. T1, T2 data from “images” category). Default is default_concat().

pymia.data.creation.traverser.default_concat(data: List[ndarray]) → ndarray

Default concatenation function used to combine all entries from a category (e.g. T1, T2 data from “images” category) in Traverser.traverse()

Parameters:

data (list) – List of numpy.ndarray entries to be concatenated.

Returns:

Concatenated entry.

Return type:

numpy.ndarray

Writer (pymia.data.creation.writer module)

class pymia.data.creation.writer.Hdf5Writer(file_path: str)

Bases: Writer

Writer class for HDF5 file type.

Parameters:

file_path (str) – The path to the dataset file to write.

close()

see Writer.close()

fill(entry: str, data, index: IndexExpression | None = None)

see Writer.fill()

open()

see Writer.open()

reserve(entry: str, shape: tuple, dtype=None)

see Writer.reserve()

write(entry: str, data, dtype=None)

see Writer.write()

class pymia.data.creation.writer.Writer

Bases: ABC

Represents the abstract dataset writer defining an interface for the writing process.

abstract close()

Close the writer.

abstract fill(entry: str, data, index: IndexExpression | None = None)

Fill parts of a reserved dataset entry.

Parameters:

• entry (str) – The dataset entry to be filled.

• data (object) – The data to write.

• index (.IndexExpression) – The slicing expression.

abstract open()

Open the writer.

abstract reserve(entry: str, shape: tuple, dtype=None)

Reserve space in the dataset for later writing.

Parameters:

• entry (str) – The dataset entry to be created.

• shape (tuple) – The shape to be reserved.

• dtype – The dtype.

abstract write(entry: str, data, dtype=None)

Create and write entry.

Parameters:

• entry (str) – The dataset entry to be written.

• data (object) – The data to write.

• dtype – The dtype.

pymia.data.creation.writer.get_writer(file_path: str) → Writer

Get the dataset writer corresponding to the file extension.

Parameters:

file_path (str) – The path of the dataset file to be written.

Returns:

Writer corresponding to dataset file extension.

Return type:

.creation.writer.Writer

pymia.data.creation.writer.writer_registry = {'.h5': <class 'pymia.data.creation.writer.Hdf5Writer'>, '.hdf5': <class 'pymia.data.creation.writer.Hdf5Writer'>}

Registry defining the mapping between file extension and Writer class. Alternative writers need to be added to this registry in order to use get_writer().