Skip to content

Medical image datasets

TorchIO offers tools to easily download publicly available datasets from different institutions and modalities.

The interface is similar to torchvision.datasets.

If you use any of them, please visit the corresponding website (linked in each description) and make sure you comply with any data usage agreement and you acknowledge the corresponding authors' publications.

If you would like to add a dataset here, please open a discussion on the GitHub repository.

CT-RATE

CtRate

CtRate

Bases: SubjectsDataset

CT-RATE dataset.

This class helps loading the CT-RATE dataset , which contains chest CT scans with associated radiology reports and abnormality labels.

The dataset must have been downloaded previously.

Parameters:

Name Type Description Default
root TypePath

Root directory where the dataset has been downloaded.

 required
split TypeSplit

Dataset split to use, either 'train' or 'validation'.

 'train'
num_subjects int | None

Optional limit on the number of subjects to load (useful for debugging). If None, all subjects in the split are loaded.

 None
report_key str

Key to use for storing radiology reports in the Subject metadata.

 'report'
sizes list[int] | None

List of image sizes (in-plane, in voxels) to include.

 None
load_fixed bool

If True, load the files with fixed spatial metadata added in this pull request . Otherwise, load the original files with incorrect spatial metadata.

 True
verify_paths bool

If True, verify that the paths to the images exist during instantiation of the dataset. This might be slow for large that are not stored locally.

 False
**kwargs

Additional arguments for SubjectsDataset.

 {}

Examples:

>>> from torchio.datasets import CtRate
>>> dataset = CtRate('/path/to/CT-RATE', sizes=[512])

from_batch(batch) classmethod

Instantiate a dataset from a batch generated by a data loader.

Parameters:

Name Type Description Default
batch dict

Dictionary generated by a data loader, containing data that can be converted to instances of [Subject][.torchio.Subject].

 required

dry_iter()

Return the internal list of subjects.

This can be used to iterate over the subjects without loading the data and applying any transforms.

Examples:

>>> names = [subject.name for subject in dataset.dry_iter()]

set_transform(transform)

Set the transform attribute.

Parameters:

Name Type Description Default
transform Callable | None

Callable object, typically an subclass of torchio.transforms.Transform.

 required

IXI

IXI

IXI

Bases: SubjectsDataset

Full IXI dataset.

Parameters:

Name Type Description Default
root TypePath

Root directory to which the dataset will be downloaded.

 required
transform

An instance of Transform.

 required
download bool

If set to True, will download the data into root.

 False
modalities Sequence[str]

List of modalities to be downloaded. They must be in ('T1', 'T2', 'PD', 'MRA', 'DTI').

 ('T1', 'T2')
Warning

The size of this dataset is multiple GB. If you set download to True, it will take some time to be downloaded if it is not already present.

Examples:

>>> import torchio as tio
>>> transforms = [
...     tio.ToCanonical(),  # to RAS
...     tio.Resample((1, 1, 1)),  # to 1 mm iso
... ]
>>> ixi_dataset = tio.datasets.IXI(
...     'path/to/ixi_root/',
...     modalities=('T1', 'T2'),
...     transform=tio.Compose(transforms),
...     download=True,
... )
>>> print('Number of subjects in dataset:', len(ixi_dataset))  # 577
>>> sample_subject = ixi_dataset[0]
>>> print('Keys in subject:', tuple(sample_subject.keys()))  # ('T1', 'T2')
>>> print('Shape of T1 data:', sample_subject['T1'].shape)  # [1, 180, 268, 268]
>>> print('Shape of T2 data:', sample_subject['T2'].shape)  # [1, 241, 257, 188]

from_batch(batch) classmethod

Instantiate a dataset from a batch generated by a data loader.

Parameters:

Name Type Description Default
batch dict

Dictionary generated by a data loader, containing data that can be converted to instances of [Subject][.torchio.Subject].

 required

dry_iter()

Return the internal list of subjects.

This can be used to iterate over the subjects without loading the data and applying any transforms.

Examples:

>>> names = [subject.name for subject in dataset.dry_iter()]

set_transform(transform)

Set the transform attribute.

Parameters:

Name Type Description Default
transform Callable | None

Callable object, typically an subclass of torchio.transforms.Transform.

 required

IXITiny

IXITiny

Bases: SubjectsDataset

This is the dataset used in the main notebook. It is a tiny version of IXI, containing 566 \(T_1\)-weighted brain MR images and their corresponding brain segmentations, all with size \(83 \times 44 \times 55\).

It can be used as a medical image MNIST.

Parameters:

Name Type Description Default
root TypePath

Root directory to which the dataset will be downloaded.

 required
transform Transform | None

An instance of Transform.

 None
download bool

If set to True, will download the data into root.

 False

from_batch(batch) classmethod

Instantiate a dataset from a batch generated by a data loader.

Parameters:

Name Type Description Default
batch dict

Dictionary generated by a data loader, containing data that can be converted to instances of [Subject][.torchio.Subject].

 required

dry_iter()

Return the internal list of subjects.

This can be used to iterate over the subjects without loading the data and applying any transforms.

Examples:

>>> names = [subject.name for subject in dataset.dry_iter()]

set_transform(transform)

Set the transform attribute.

Parameters:

Name Type Description Default
transform Callable | None

Callable object, typically an subclass of torchio.transforms.Transform.

 required

EPISURG

EPISURG

EPISURG

Bases: SubjectsDataset

EPISURG is a clinical dataset of \(T_1\)-weighted MRI from 430 epileptic patients who underwent resective brain surgery at the National Hospital of Neurology and Neurosurgery (Queen Square, London, United Kingdom) between 1990 and 2018.

The dataset comprises 430 postoperative MRI. The corresponding preoperative MRI is present for 268 subjects.

Three human raters segmented the resection cavity on partially overlapping subsets of EPISURG.

If you use this dataset for your research, you agree with the Data use agreement presented at the EPISURG entry on the UCL Research Data Repository and you must cite the corresponding publications.

Parameters:

Name Type Description Default
root TypePath

Root directory to which the dataset will be downloaded.

 required
transform Transform | None

An instance of Transform.

 None
download bool

If set to True, will download the data into root.

 False
Warning

The size of this dataset is multiple GB. If you set download to True, it will take some time to be downloaded if it is not already present.

from_batch(batch) classmethod

Instantiate a dataset from a batch generated by a data loader.

Parameters:

Name Type Description Default
batch dict

Dictionary generated by a data loader, containing data that can be converted to instances of [Subject][.torchio.Subject].

 required

dry_iter()

Return the internal list of subjects.

This can be used to iterate over the subjects without loading the data and applying any transforms.

Examples:

>>> names = [subject.name for subject in dataset.dry_iter()]

set_transform(transform)

Set the transform attribute.

Parameters:

Name Type Description Default
transform Callable | None

Callable object, typically an subclass of torchio.transforms.Transform.

 required

get_labeled()

Get dataset from subjects with manual annotations.

get_unlabeled()

Get dataset from subjects without manual annotations.

get_paired()

Get dataset from subjects with pre- and post-op MRI.

Kaggle datasets

RSNAMICCAI

RSNAMICCAI

Bases: SubjectsDataset

RSNA-MICCAI Brain Tumor Radiogenomic Classification challenge dataset.

This is a helper class for the dataset used in the RSNA-MICCAI Brain Tumor Radiogenomic Classification challenge hosted on kaggle . The dataset must be downloaded before instantiating this class (as opposed to, e.g., torchio.datasets.IXI).

This kaggle kernel includes a usage example including preprocessing of all the scans.

If you reference or use the dataset in any form, include the following citation:

U.Baid, et al., "The RSNA-ASNR-MICCAI BraTS 2021 Benchmark on Brain Tumor Segmentation and Radiogenomic Classification", arXiv:2107.02314, 2021.

Parameters:

Name Type Description Default
root_dir TypePath

Directory containing the dataset (train directory, test directory, etc.).

 required
train bool

If True, the train set will be used. Otherwise the test set will be used.

 True
ignore_empty bool

If True, the three subjects flagged as "presenting issues" (empty images) by the challenge organizers will be ignored. The subject IDs are 00109, 00123 and 00709.

 True

Examples:

>>> import torchio as tio
>>> from subprocess import call
>>> call('kaggle competitions download -c rsna-miccai-brain-tumor-radiogenomic-classification'.split())
>>> root_dir = 'rsna-miccai-brain-tumor-radiogenomic-classification'
>>> train_set = tio.datasets.RSNAMICCAI(root_dir, train=True)
>>> test_set = tio.datasets.RSNAMICCAI(root_dir, train=False)
>>> len(train_set), len(test_set)
(582, 87)

from_batch(batch) classmethod

Instantiate a dataset from a batch generated by a data loader.

Parameters:

Name Type Description Default
batch dict

Dictionary generated by a data loader, containing data that can be converted to instances of [Subject][.torchio.Subject].

 required

dry_iter()

Return the internal list of subjects.

This can be used to iterate over the subjects without loading the data and applying any transforms.

Examples:

>>> names = [subject.name for subject in dataset.dry_iter()]

set_transform(transform)

Set the transform attribute.

Parameters:

Name Type Description Default
transform Callable | None

Callable object, typically an subclass of torchio.transforms.Transform.

 required

RSNACervicalSpineFracture

RSNACervicalSpineFracture

Bases: SubjectsDataset

RSNA 2022 Cervical Spine Fracture Detection dataset.

This is a helper class for the dataset used in the RSNA 2022 Cervical Spine Fracture Detection hosted on kaggle . The dataset must be downloaded before instantiating this class.

from_batch(batch) classmethod

Instantiate a dataset from a batch generated by a data loader.

Parameters:

Name Type Description Default
batch dict

Dictionary generated by a data loader, containing data that can be converted to instances of [Subject][.torchio.Subject].

 required

dry_iter()

Return the internal list of subjects.

This can be used to iterate over the subjects without loading the data and applying any transforms.

Examples:

>>> names = [subject.name for subject in dataset.dry_iter()]

set_transform(transform)

Set the transform attribute.

Parameters:

Name Type Description Default
transform Callable | None

Callable object, typically an subclass of torchio.transforms.Transform.

 required

MNI

ICBM2009CNonlinearSymmetric

ICBM2009CNonlinearSymmetric

Bases: SubjectMNI

ICBM template.

More information can be found in the website .

ICBM 2009c Nonlinear Symmetric

Parameters:

Name Type Description Default
load_4d_tissues bool

If True, the tissue probability maps will be loaded together into a 4D image. Otherwise, they will be loaded into independent images.

 True

Examples:

>>> import torchio as tio
>>> icbm = tio.datasets.ICBM2009CNonlinearSymmetric()
>>> icbm
ICBM2009CNonlinearSymmetric(Keys: ('t1', 'eyes', 'face', 'brain', 't2', 'pd', 'tissues'); images: 7)
>>> icbm = tio.datasets.ICBM2009CNonlinearSymmetric(load_4d_tissues=False)
>>> icbm
ICBM2009CNonlinearSymmetric(Keys: ('t1', 'eyes', 'face', 'brain', 't2', 'pd', 'gm', 'wm', 'csf'); images: 9)

shape property

Return shape of first image in subject.

Consistency of shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.shape
(1, 181, 217, 181)

spatial_shape property

Return spatial shape of first image in subject.

Consistency of spatial shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.spatial_shape
(181, 217, 181)

spacing property

Return spacing of first image in subject.

Consistency of spacings across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Slicer()
>>> colin.spacing
(1.0, 1.0, 1.2999954223632812)

get_inverse_transform(warn=True, ignore_intensity=False, image_interpolation=None)

Get a reversed list of the inverses of the applied transforms.

Parameters:

Name Type Description Default
warn bool

Issue a warning if some transforms are not invertible.

 True
ignore_intensity bool

If True, all instances of [IntensityTransform][torchio.transforms.intensity_transform.IntensityTransform] will be ignored.

 False
image_interpolation str | None

Modify interpolation for scalar images inside transforms that perform resampling.

 None

apply_inverse_transform(**kwargs)

Apply the inverse of all applied transforms, in reverse order.

Parameters:

Name Type Description Default
**kwargs

Keyword arguments passed on to get_inverse_transform().

 {}

check_consistent_attribute(attribute, relative_tolerance=1e-06, absolute_tolerance=1e-06, message=None)

Check for consistency of an attribute across all images.

Parameters:

Name Type Description Default
attribute str

Name of the image attribute to check

 required
relative_tolerance float

Relative tolerance for numpy.allclose()

 1e-06
absolute_tolerance float

Absolute tolerance for numpy.allclose()

 1e-06

Examples:

>>> import numpy as np
>>> import torch
>>> import torchio as tio
>>> scalars = torch.randn(1, 512, 512, 100)
>>> mask = torch.tensor(scalars > 0).type(torch.int16)
>>> af1 = np.eye([0.8, 0.8, 2.50000000000001, 1])
>>> af2 = np.eye([0.8, 0.8, 2.49999999999999, 1])  # small difference here (e.g. due to different reader)
>>> subject = tio.Subject(
...   image = tio.ScalarImage(tensor=scalars, affine=af1),
...   mask = tio.LabelMap(tensor=mask, affine=af2)
... )
>>> subject.check_consistent_attribute('spacing')  # no error as tolerances are > 0
Note

To check that all values for a specific attribute are close between all images in the subject, numpy.allclose() is used. This function returns True if \(|a_i - b_i| \leq t_{abs} + t_{rel} * |b_i|\), where \(a_i\) and \(b_i\) are the \(i\)-th element of the same attribute of two images being compared, \(t_{abs}\) is the absolute_tolerance and \(t_{rel}\) is the relative_tolerance.

get_image(image_name)

Get a single image by its name.

load()

Load images in subject on RAM.

unload()

Unload images in subject.

add_image(image, image_name)

Add an image to the subject instance.

remove_image(image_name)

Remove an image from the subject instance.

plot(return_fig=False, **kwargs)

Plot images using matplotlib.

Parameters:

Name Type Description Default
return_fig bool

If True, return the figure instead of showing it.

 False
**kwargs

Keyword arguments that will be passed on to plot().

 {}

Colin27

Colin27

Bases: SubjectMNI

shape property

Return shape of first image in subject.

Consistency of shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.shape
(1, 181, 217, 181)

spatial_shape property

Return spatial shape of first image in subject.

Consistency of spatial shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.spatial_shape
(181, 217, 181)

spacing property

Return spacing of first image in subject.

Consistency of spacings across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Slicer()
>>> colin.spacing
(1.0, 1.0, 1.2999954223632812)

NAME_TO_LABEL = {name: label for label, name in (TISSUES_2008.items())} class-attribute instance-attribute

Colin27 MNI template.

More information can be found in the website of the 1998 and 2008 versions.

MNI Colin 27 2008 version

Parameters:

Name Type Description Default
version

Template year. It can be 1998 or 2008.

 required
Warning

The resolution of the 2008 version is quite high. The subject instance will contain four images of size \(362 \times 434 \times 362\), therefore applying a transform to it might take longer than expected.

Examples:

>>> import torchio as tio
>>> colin_1998 = tio.datasets.Colin27(version=1998)
>>> colin_1998
Colin27(Keys: ('t1', 'head', 'brain'); images: 3)
>>> colin_1998.load()
>>> colin_1998.t1
ScalarImage(shape: (1, 181, 217, 181); spacing: (1.00, 1.00, 1.00); orientation: RAS+; memory: 27.1 MiB; type: intensity)
>>>
>>> colin_2008 = tio.datasets.Colin27(version=2008)
>>> colin_2008
Colin27(Keys: ('t1', 't2', 'pd', 'cls'); images: 4)
>>> colin_2008.load()
>>> colin_2008.t1
ScalarImage(shape: (1, 362, 434, 362); spacing: (0.50, 0.50, 0.50); orientation: RAS+; memory: 217.0 MiB; type: intensity)

get_inverse_transform(warn=True, ignore_intensity=False, image_interpolation=None)

Get a reversed list of the inverses of the applied transforms.

Parameters:

Name Type Description Default
warn bool

Issue a warning if some transforms are not invertible.

 True
ignore_intensity bool

If True, all instances of [IntensityTransform][torchio.transforms.intensity_transform.IntensityTransform] will be ignored.

 False
image_interpolation str | None

Modify interpolation for scalar images inside transforms that perform resampling.

 None

apply_inverse_transform(**kwargs)

Apply the inverse of all applied transforms, in reverse order.

Parameters:

Name Type Description Default
**kwargs

Keyword arguments passed on to get_inverse_transform().

 {}

check_consistent_attribute(attribute, relative_tolerance=1e-06, absolute_tolerance=1e-06, message=None)

Check for consistency of an attribute across all images.

Parameters:

Name Type Description Default
attribute str

Name of the image attribute to check

 required
relative_tolerance float

Relative tolerance for numpy.allclose()

 1e-06
absolute_tolerance float

Absolute tolerance for numpy.allclose()

 1e-06

Examples:

>>> import numpy as np
>>> import torch
>>> import torchio as tio
>>> scalars = torch.randn(1, 512, 512, 100)
>>> mask = torch.tensor(scalars > 0).type(torch.int16)
>>> af1 = np.eye([0.8, 0.8, 2.50000000000001, 1])
>>> af2 = np.eye([0.8, 0.8, 2.49999999999999, 1])  # small difference here (e.g. due to different reader)
>>> subject = tio.Subject(
...   image = tio.ScalarImage(tensor=scalars, affine=af1),
...   mask = tio.LabelMap(tensor=mask, affine=af2)
... )
>>> subject.check_consistent_attribute('spacing')  # no error as tolerances are > 0
Note

To check that all values for a specific attribute are close between all images in the subject, numpy.allclose() is used. This function returns True if \(|a_i - b_i| \leq t_{abs} + t_{rel} * |b_i|\), where \(a_i\) and \(b_i\) are the \(i\)-th element of the same attribute of two images being compared, \(t_{abs}\) is the absolute_tolerance and \(t_{rel}\) is the relative_tolerance.

get_image(image_name)

Get a single image by its name.

load()

Load images in subject on RAM.

unload()

Unload images in subject.

add_image(image, image_name)

Add an image to the subject instance.

remove_image(image_name)

Remove an image from the subject instance.

plot(return_fig=False, **kwargs)

Plot images using matplotlib.

Parameters:

Name Type Description Default
return_fig bool

If True, return the figure instead of showing it.

 False
**kwargs

Keyword arguments that will be passed on to plot().

 {}

Pediatric

Pediatric

Bases: SubjectMNI

MNI pediatric atlases.

See the MNI website for more information.

Pediatric MNI template

Parameters:

Name Type Description Default
years

Tuple of 2 ages. Possible values are: (4.5, 18.5), (4.5, 8.5), (7, 11), (7.5, 13.5), (10, 14) and (13, 18.5).

 required
symmetric

If True, the left-right symmetric templates will be used. Else, the asymmetric (natural) templates will be used.

 False

shape property

Return shape of first image in subject.

Consistency of shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.shape
(1, 181, 217, 181)

spatial_shape property

Return spatial shape of first image in subject.

Consistency of spatial shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.spatial_shape
(181, 217, 181)

spacing property

Return spacing of first image in subject.

Consistency of spacings across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Slicer()
>>> colin.spacing
(1.0, 1.0, 1.2999954223632812)

get_inverse_transform(warn=True, ignore_intensity=False, image_interpolation=None)

Get a reversed list of the inverses of the applied transforms.

Parameters:

Name Type Description Default
warn bool

Issue a warning if some transforms are not invertible.

 True
ignore_intensity bool

If True, all instances of [IntensityTransform][torchio.transforms.intensity_transform.IntensityTransform] will be ignored.

 False
image_interpolation str | None

Modify interpolation for scalar images inside transforms that perform resampling.

 None

apply_inverse_transform(**kwargs)

Apply the inverse of all applied transforms, in reverse order.

Parameters:

Name Type Description Default
**kwargs

Keyword arguments passed on to get_inverse_transform().

 {}

check_consistent_attribute(attribute, relative_tolerance=1e-06, absolute_tolerance=1e-06, message=None)

Check for consistency of an attribute across all images.

Parameters:

Name Type Description Default
attribute str

Name of the image attribute to check

 required
relative_tolerance float

Relative tolerance for numpy.allclose()

 1e-06
absolute_tolerance float

Absolute tolerance for numpy.allclose()

 1e-06

Examples:

>>> import numpy as np
>>> import torch
>>> import torchio as tio
>>> scalars = torch.randn(1, 512, 512, 100)
>>> mask = torch.tensor(scalars > 0).type(torch.int16)
>>> af1 = np.eye([0.8, 0.8, 2.50000000000001, 1])
>>> af2 = np.eye([0.8, 0.8, 2.49999999999999, 1])  # small difference here (e.g. due to different reader)
>>> subject = tio.Subject(
...   image = tio.ScalarImage(tensor=scalars, affine=af1),
...   mask = tio.LabelMap(tensor=mask, affine=af2)
... )
>>> subject.check_consistent_attribute('spacing')  # no error as tolerances are > 0
Note

To check that all values for a specific attribute are close between all images in the subject, numpy.allclose() is used. This function returns True if \(|a_i - b_i| \leq t_{abs} + t_{rel} * |b_i|\), where \(a_i\) and \(b_i\) are the \(i\)-th element of the same attribute of two images being compared, \(t_{abs}\) is the absolute_tolerance and \(t_{rel}\) is the relative_tolerance.

get_image(image_name)

Get a single image by its name.

load()

Load images in subject on RAM.

unload()

Unload images in subject.

add_image(image, image_name)

Add an image to the subject instance.

remove_image(image_name)

Remove an image from the subject instance.

plot(return_fig=False, **kwargs)

Plot images using matplotlib.

Parameters:

Name Type Description Default
return_fig bool

If True, return the figure instead of showing it.

 False
**kwargs

Keyword arguments that will be passed on to plot().

 {}

Sheep

Sheep

Bases: SubjectMNI

shape property

Return shape of first image in subject.

Consistency of shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.shape
(1, 181, 217, 181)

spatial_shape property

Return spatial shape of first image in subject.

Consistency of spatial shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.spatial_shape
(181, 217, 181)

spacing property

Return spacing of first image in subject.

Consistency of spacings across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Slicer()
>>> colin.spacing
(1.0, 1.0, 1.2999954223632812)

get_inverse_transform(warn=True, ignore_intensity=False, image_interpolation=None)

Get a reversed list of the inverses of the applied transforms.

Parameters:

Name Type Description Default
warn bool

Issue a warning if some transforms are not invertible.

 True
ignore_intensity bool

If True, all instances of [IntensityTransform][torchio.transforms.intensity_transform.IntensityTransform] will be ignored.

 False
image_interpolation str | None

Modify interpolation for scalar images inside transforms that perform resampling.

 None

apply_inverse_transform(**kwargs)

Apply the inverse of all applied transforms, in reverse order.

Parameters:

Name Type Description Default
**kwargs

Keyword arguments passed on to get_inverse_transform().

 {}

check_consistent_attribute(attribute, relative_tolerance=1e-06, absolute_tolerance=1e-06, message=None)

Check for consistency of an attribute across all images.

Parameters:

Name Type Description Default
attribute str

Name of the image attribute to check

 required
relative_tolerance float

Relative tolerance for numpy.allclose()

 1e-06
absolute_tolerance float

Absolute tolerance for numpy.allclose()

 1e-06

Examples:

>>> import numpy as np
>>> import torch
>>> import torchio as tio
>>> scalars = torch.randn(1, 512, 512, 100)
>>> mask = torch.tensor(scalars > 0).type(torch.int16)
>>> af1 = np.eye([0.8, 0.8, 2.50000000000001, 1])
>>> af2 = np.eye([0.8, 0.8, 2.49999999999999, 1])  # small difference here (e.g. due to different reader)
>>> subject = tio.Subject(
...   image = tio.ScalarImage(tensor=scalars, affine=af1),
...   mask = tio.LabelMap(tensor=mask, affine=af2)
... )
>>> subject.check_consistent_attribute('spacing')  # no error as tolerances are > 0
Note

To check that all values for a specific attribute are close between all images in the subject, numpy.allclose() is used. This function returns True if \(|a_i - b_i| \leq t_{abs} + t_{rel} * |b_i|\), where \(a_i\) and \(b_i\) are the \(i\)-th element of the same attribute of two images being compared, \(t_{abs}\) is the absolute_tolerance and \(t_{rel}\) is the relative_tolerance.

get_image(image_name)

Get a single image by its name.

load()

Load images in subject on RAM.

unload()

Unload images in subject.

add_image(image, image_name)

Add an image to the subject instance.

remove_image(image_name)

Remove an image from the subject instance.

plot(return_fig=False, **kwargs)

Plot images using matplotlib.

Parameters:

Name Type Description Default
return_fig bool

If True, return the figure instead of showing it.

 False
**kwargs

Keyword arguments that will be passed on to plot().

 {}

BITE3

BITE3

Bases: BITE

Pre- and post-resection MR images in BITE.

The goal of BITE is to share in vivo medical images of patients wtith brain tumors to facilitate the development and validation of new image processing algorithms.

Please check the BITE website for more information and acknowledgments instructions.

Parameters:

Name Type Description Default
root TypePath

Root directory to which the dataset will be downloaded.

 required
transform Transform | None

An instance of Transform.

 None
download bool

If set to True, will download the data into root.

 False

from_batch(batch) classmethod

Instantiate a dataset from a batch generated by a data loader.

Parameters:

Name Type Description Default
batch dict

Dictionary generated by a data loader, containing data that can be converted to instances of [Subject][.torchio.Subject].

 required

dry_iter()

Return the internal list of subjects.

This can be used to iterate over the subjects without loading the data and applying any transforms.

Examples:

>>> names = [subject.name for subject in dataset.dry_iter()]

set_transform(transform)

Set the transform attribute.

Parameters:

Name Type Description Default
transform Callable | None

Callable object, typically an subclass of torchio.transforms.Transform.

 required

ITK-SNAP

BrainTumor

BrainTumor

Bases: SubjectITKSNAP

shape property

Return shape of first image in subject.

Consistency of shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.shape
(1, 181, 217, 181)

spatial_shape property

Return spatial shape of first image in subject.

Consistency of spatial shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.spatial_shape
(181, 217, 181)

spacing property

Return spacing of first image in subject.

Consistency of spacings across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Slicer()
>>> colin.spacing
(1.0, 1.0, 1.2999954223632812)

get_inverse_transform(warn=True, ignore_intensity=False, image_interpolation=None)

Get a reversed list of the inverses of the applied transforms.

Parameters:

Name Type Description Default
warn bool

Issue a warning if some transforms are not invertible.

 True
ignore_intensity bool

If True, all instances of [IntensityTransform][torchio.transforms.intensity_transform.IntensityTransform] will be ignored.

 False
image_interpolation str | None

Modify interpolation for scalar images inside transforms that perform resampling.

 None

apply_inverse_transform(**kwargs)

Apply the inverse of all applied transforms, in reverse order.

Parameters:

Name Type Description Default
**kwargs

Keyword arguments passed on to get_inverse_transform().

 {}

check_consistent_attribute(attribute, relative_tolerance=1e-06, absolute_tolerance=1e-06, message=None)

Check for consistency of an attribute across all images.

Parameters:

Name Type Description Default
attribute str

Name of the image attribute to check

 required
relative_tolerance float

Relative tolerance for numpy.allclose()

 1e-06
absolute_tolerance float

Absolute tolerance for numpy.allclose()

 1e-06

Examples:

>>> import numpy as np
>>> import torch
>>> import torchio as tio
>>> scalars = torch.randn(1, 512, 512, 100)
>>> mask = torch.tensor(scalars > 0).type(torch.int16)
>>> af1 = np.eye([0.8, 0.8, 2.50000000000001, 1])
>>> af2 = np.eye([0.8, 0.8, 2.49999999999999, 1])  # small difference here (e.g. due to different reader)
>>> subject = tio.Subject(
...   image = tio.ScalarImage(tensor=scalars, affine=af1),
...   mask = tio.LabelMap(tensor=mask, affine=af2)
... )
>>> subject.check_consistent_attribute('spacing')  # no error as tolerances are > 0
Note

To check that all values for a specific attribute are close between all images in the subject, numpy.allclose() is used. This function returns True if \(|a_i - b_i| \leq t_{abs} + t_{rel} * |b_i|\), where \(a_i\) and \(b_i\) are the \(i\)-th element of the same attribute of two images being compared, \(t_{abs}\) is the absolute_tolerance and \(t_{rel}\) is the relative_tolerance.

get_image(image_name)

Get a single image by its name.

load()

Load images in subject on RAM.

unload()

Unload images in subject.

add_image(image, image_name)

Add an image to the subject instance.

remove_image(image_name)

Remove an image from the subject instance.

plot(return_fig=False, **kwargs)

Plot images using matplotlib.

Parameters:

Name Type Description Default
return_fig bool

If True, return the figure instead of showing it.

 False
**kwargs

Keyword arguments that will be passed on to plot().

 {}

T1T2

T1T2

Bases: SubjectITKSNAP

shape property

Return shape of first image in subject.

Consistency of shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.shape
(1, 181, 217, 181)

spatial_shape property

Return spatial shape of first image in subject.

Consistency of spatial shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.spatial_shape
(181, 217, 181)

spacing property

Return spacing of first image in subject.

Consistency of spacings across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Slicer()
>>> colin.spacing
(1.0, 1.0, 1.2999954223632812)

get_inverse_transform(warn=True, ignore_intensity=False, image_interpolation=None)

Get a reversed list of the inverses of the applied transforms.

Parameters:

Name Type Description Default
warn bool

Issue a warning if some transforms are not invertible.

 True
ignore_intensity bool

If True, all instances of [IntensityTransform][torchio.transforms.intensity_transform.IntensityTransform] will be ignored.

 False
image_interpolation str | None

Modify interpolation for scalar images inside transforms that perform resampling.

 None

apply_inverse_transform(**kwargs)

Apply the inverse of all applied transforms, in reverse order.

Parameters:

Name Type Description Default
**kwargs

Keyword arguments passed on to get_inverse_transform().

 {}

check_consistent_attribute(attribute, relative_tolerance=1e-06, absolute_tolerance=1e-06, message=None)

Check for consistency of an attribute across all images.

Parameters:

Name Type Description Default
attribute str

Name of the image attribute to check

 required
relative_tolerance float

Relative tolerance for numpy.allclose()

 1e-06
absolute_tolerance float

Absolute tolerance for numpy.allclose()

 1e-06

Examples:

>>> import numpy as np
>>> import torch
>>> import torchio as tio
>>> scalars = torch.randn(1, 512, 512, 100)
>>> mask = torch.tensor(scalars > 0).type(torch.int16)
>>> af1 = np.eye([0.8, 0.8, 2.50000000000001, 1])
>>> af2 = np.eye([0.8, 0.8, 2.49999999999999, 1])  # small difference here (e.g. due to different reader)
>>> subject = tio.Subject(
...   image = tio.ScalarImage(tensor=scalars, affine=af1),
...   mask = tio.LabelMap(tensor=mask, affine=af2)
... )
>>> subject.check_consistent_attribute('spacing')  # no error as tolerances are > 0
Note

To check that all values for a specific attribute are close between all images in the subject, numpy.allclose() is used. This function returns True if \(|a_i - b_i| \leq t_{abs} + t_{rel} * |b_i|\), where \(a_i\) and \(b_i\) are the \(i\)-th element of the same attribute of two images being compared, \(t_{abs}\) is the absolute_tolerance and \(t_{rel}\) is the relative_tolerance.

get_image(image_name)

Get a single image by its name.

load()

Load images in subject on RAM.

unload()

Unload images in subject.

add_image(image, image_name)

Add an image to the subject instance.

remove_image(image_name)

Remove an image from the subject instance.

plot(return_fig=False, **kwargs)

Plot images using matplotlib.

Parameters:

Name Type Description Default
return_fig bool

If True, return the figure instead of showing it.

 False
**kwargs

Keyword arguments that will be passed on to plot().

 {}

AorticValve

AorticValve

Bases: SubjectITKSNAP

shape property

Return shape of first image in subject.

Consistency of shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.shape
(1, 181, 217, 181)

spatial_shape property

Return spatial shape of first image in subject.

Consistency of spatial shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.spatial_shape
(181, 217, 181)

spacing property

Return spacing of first image in subject.

Consistency of spacings across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Slicer()
>>> colin.spacing
(1.0, 1.0, 1.2999954223632812)

get_inverse_transform(warn=True, ignore_intensity=False, image_interpolation=None)

Get a reversed list of the inverses of the applied transforms.

Parameters:

Name Type Description Default
warn bool

Issue a warning if some transforms are not invertible.

 True
ignore_intensity bool

If True, all instances of [IntensityTransform][torchio.transforms.intensity_transform.IntensityTransform] will be ignored.

 False
image_interpolation str | None

Modify interpolation for scalar images inside transforms that perform resampling.

 None

apply_inverse_transform(**kwargs)

Apply the inverse of all applied transforms, in reverse order.

Parameters:

Name Type Description Default
**kwargs

Keyword arguments passed on to get_inverse_transform().

 {}

check_consistent_attribute(attribute, relative_tolerance=1e-06, absolute_tolerance=1e-06, message=None)

Check for consistency of an attribute across all images.

Parameters:

Name Type Description Default
attribute str

Name of the image attribute to check

 required
relative_tolerance float

Relative tolerance for numpy.allclose()

 1e-06
absolute_tolerance float

Absolute tolerance for numpy.allclose()

 1e-06

Examples:

>>> import numpy as np
>>> import torch
>>> import torchio as tio
>>> scalars = torch.randn(1, 512, 512, 100)
>>> mask = torch.tensor(scalars > 0).type(torch.int16)
>>> af1 = np.eye([0.8, 0.8, 2.50000000000001, 1])
>>> af2 = np.eye([0.8, 0.8, 2.49999999999999, 1])  # small difference here (e.g. due to different reader)
>>> subject = tio.Subject(
...   image = tio.ScalarImage(tensor=scalars, affine=af1),
...   mask = tio.LabelMap(tensor=mask, affine=af2)
... )
>>> subject.check_consistent_attribute('spacing')  # no error as tolerances are > 0
Note

To check that all values for a specific attribute are close between all images in the subject, numpy.allclose() is used. This function returns True if \(|a_i - b_i| \leq t_{abs} + t_{rel} * |b_i|\), where \(a_i\) and \(b_i\) are the \(i\)-th element of the same attribute of two images being compared, \(t_{abs}\) is the absolute_tolerance and \(t_{rel}\) is the relative_tolerance.

get_image(image_name)

Get a single image by its name.

load()

Load images in subject on RAM.

unload()

Unload images in subject.

add_image(image, image_name)

Add an image to the subject instance.

remove_image(image_name)

Remove an image from the subject instance.

plot(return_fig=False, **kwargs)

Plot images using matplotlib.

Parameters:

Name Type Description Default
return_fig bool

If True, return the figure instead of showing it.

 False
**kwargs

Keyword arguments that will be passed on to plot().

 {}

3D Slicer

Slicer

Slicer

Bases: Subject

Sample data provided by 3D Slicer .

See the Slicer wiki for more information.

For information about licensing and permissions, check the Sample Data module .

Parameters:

Name Type Description Default
name

One of the keys in torchio.datasets.slicer.URLS_DICT.

 'MRHead'

shape property

Return shape of first image in subject.

Consistency of shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.shape
(1, 181, 217, 181)

spatial_shape property

Return spatial shape of first image in subject.

Consistency of spatial shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.spatial_shape
(181, 217, 181)

spacing property

Return spacing of first image in subject.

Consistency of spacings across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Slicer()
>>> colin.spacing
(1.0, 1.0, 1.2999954223632812)

get_inverse_transform(warn=True, ignore_intensity=False, image_interpolation=None)

Get a reversed list of the inverses of the applied transforms.

Parameters:

Name Type Description Default
warn bool

Issue a warning if some transforms are not invertible.

 True
ignore_intensity bool

If True, all instances of [IntensityTransform][torchio.transforms.intensity_transform.IntensityTransform] will be ignored.

 False
image_interpolation str | None

Modify interpolation for scalar images inside transforms that perform resampling.

 None

apply_inverse_transform(**kwargs)

Apply the inverse of all applied transforms, in reverse order.

Parameters:

Name Type Description Default
**kwargs

Keyword arguments passed on to get_inverse_transform().

 {}

check_consistent_attribute(attribute, relative_tolerance=1e-06, absolute_tolerance=1e-06, message=None)

Check for consistency of an attribute across all images.

Parameters:

Name Type Description Default
attribute str

Name of the image attribute to check

 required
relative_tolerance float

Relative tolerance for numpy.allclose()

 1e-06
absolute_tolerance float

Absolute tolerance for numpy.allclose()

 1e-06

Examples:

>>> import numpy as np
>>> import torch
>>> import torchio as tio
>>> scalars = torch.randn(1, 512, 512, 100)
>>> mask = torch.tensor(scalars > 0).type(torch.int16)
>>> af1 = np.eye([0.8, 0.8, 2.50000000000001, 1])
>>> af2 = np.eye([0.8, 0.8, 2.49999999999999, 1])  # small difference here (e.g. due to different reader)
>>> subject = tio.Subject(
...   image = tio.ScalarImage(tensor=scalars, affine=af1),
...   mask = tio.LabelMap(tensor=mask, affine=af2)
... )
>>> subject.check_consistent_attribute('spacing')  # no error as tolerances are > 0
Note

To check that all values for a specific attribute are close between all images in the subject, numpy.allclose() is used. This function returns True if \(|a_i - b_i| \leq t_{abs} + t_{rel} * |b_i|\), where \(a_i\) and \(b_i\) are the \(i\)-th element of the same attribute of two images being compared, \(t_{abs}\) is the absolute_tolerance and \(t_{rel}\) is the relative_tolerance.

get_image(image_name)

Get a single image by its name.

load()

Load images in subject on RAM.

unload()

Unload images in subject.

add_image(image, image_name)

Add an image to the subject instance.

remove_image(image_name)

Remove an image from the subject instance.

plot(return_fig=False, **kwargs)

Plot images using matplotlib.

Parameters:

Name Type Description Default
return_fig bool

If True, return the figure instead of showing it.

 False
**kwargs

Keyword arguments that will be passed on to plot().

 {}

FPG

FPG

FPG

Bases: Subject

3T \(T_1\)-weighted brain MRI and corresponding parcellation.

Parameters:

Name Type Description Default
load_all bool

If True, three more images will be loaded: a \(T_2\)-weighted MRI, a diffusion MRI and a functional MRI.

 False

shape property

Return shape of first image in subject.

Consistency of shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.shape
(1, 181, 217, 181)

spatial_shape property

Return spatial shape of first image in subject.

Consistency of spatial shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.spatial_shape
(181, 217, 181)

spacing property

Return spacing of first image in subject.

Consistency of spacings across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Slicer()
>>> colin.spacing
(1.0, 1.0, 1.2999954223632812)

get_inverse_transform(warn=True, ignore_intensity=False, image_interpolation=None)

Get a reversed list of the inverses of the applied transforms.

Parameters:

Name Type Description Default
warn bool

Issue a warning if some transforms are not invertible.

 True
ignore_intensity bool

If True, all instances of [IntensityTransform][torchio.transforms.intensity_transform.IntensityTransform] will be ignored.

 False
image_interpolation str | None

Modify interpolation for scalar images inside transforms that perform resampling.

 None

apply_inverse_transform(**kwargs)

Apply the inverse of all applied transforms, in reverse order.

Parameters:

Name Type Description Default
**kwargs

Keyword arguments passed on to get_inverse_transform().

 {}

check_consistent_attribute(attribute, relative_tolerance=1e-06, absolute_tolerance=1e-06, message=None)

Check for consistency of an attribute across all images.

Parameters:

Name Type Description Default
attribute str

Name of the image attribute to check

 required
relative_tolerance float

Relative tolerance for numpy.allclose()

 1e-06
absolute_tolerance float

Absolute tolerance for numpy.allclose()

 1e-06

Examples:

>>> import numpy as np
>>> import torch
>>> import torchio as tio
>>> scalars = torch.randn(1, 512, 512, 100)
>>> mask = torch.tensor(scalars > 0).type(torch.int16)
>>> af1 = np.eye([0.8, 0.8, 2.50000000000001, 1])
>>> af2 = np.eye([0.8, 0.8, 2.49999999999999, 1])  # small difference here (e.g. due to different reader)
>>> subject = tio.Subject(
...   image = tio.ScalarImage(tensor=scalars, affine=af1),
...   mask = tio.LabelMap(tensor=mask, affine=af2)
... )
>>> subject.check_consistent_attribute('spacing')  # no error as tolerances are > 0
Note

To check that all values for a specific attribute are close between all images in the subject, numpy.allclose() is used. This function returns True if \(|a_i - b_i| \leq t_{abs} + t_{rel} * |b_i|\), where \(a_i\) and \(b_i\) are the \(i\)-th element of the same attribute of two images being compared, \(t_{abs}\) is the absolute_tolerance and \(t_{rel}\) is the relative_tolerance.

get_image(image_name)

Get a single image by its name.

load()

Load images in subject on RAM.

unload()

Unload images in subject.

add_image(image, image_name)

Add an image to the subject instance.

remove_image(image_name)

Remove an image from the subject instance.

MedMNIST

OrganMNIST3D

OrganMNIST3D

Bases: MedMNIST

from_batch(batch) classmethod

Instantiate a dataset from a batch generated by a data loader.

Parameters:

Name Type Description Default
batch dict

Dictionary generated by a data loader, containing data that can be converted to instances of [Subject][.torchio.Subject].

 required

dry_iter()

Return the internal list of subjects.

This can be used to iterate over the subjects without loading the data and applying any transforms.

Examples:

>>> names = [subject.name for subject in dataset.dry_iter()]

set_transform(transform)

Set the transform attribute.

Parameters:

Name Type Description Default
transform Callable | None

Callable object, typically an subclass of torchio.transforms.Transform.

 required

NoduleMNIST3D

NoduleMNIST3D

Bases: MedMNIST

from_batch(batch) classmethod

Instantiate a dataset from a batch generated by a data loader.

Parameters:

Name Type Description Default
batch dict

Dictionary generated by a data loader, containing data that can be converted to instances of [Subject][.torchio.Subject].

 required

dry_iter()

Return the internal list of subjects.

This can be used to iterate over the subjects without loading the data and applying any transforms.

Examples:

>>> names = [subject.name for subject in dataset.dry_iter()]

set_transform(transform)

Set the transform attribute.

Parameters:

Name Type Description Default
transform Callable | None

Callable object, typically an subclass of torchio.transforms.Transform.

 required

AdrenalMNIST3D

AdrenalMNIST3D

Bases: MedMNIST

from_batch(batch) classmethod

Instantiate a dataset from a batch generated by a data loader.

Parameters:

Name Type Description Default
batch dict

Dictionary generated by a data loader, containing data that can be converted to instances of [Subject][.torchio.Subject].

 required

dry_iter()

Return the internal list of subjects.

This can be used to iterate over the subjects without loading the data and applying any transforms.

Examples:

>>> names = [subject.name for subject in dataset.dry_iter()]

set_transform(transform)

Set the transform attribute.

Parameters:

Name Type Description Default
transform Callable | None

Callable object, typically an subclass of torchio.transforms.Transform.

 required

FractureMNIST3D

FractureMNIST3D

Bases: MedMNIST

from_batch(batch) classmethod

Instantiate a dataset from a batch generated by a data loader.

Parameters:

Name Type Description Default
batch dict

Dictionary generated by a data loader, containing data that can be converted to instances of [Subject][.torchio.Subject].

 required

dry_iter()

Return the internal list of subjects.

This can be used to iterate over the subjects without loading the data and applying any transforms.

Examples:

>>> names = [subject.name for subject in dataset.dry_iter()]

set_transform(transform)

Set the transform attribute.

Parameters:

Name Type Description Default
transform Callable | None

Callable object, typically an subclass of torchio.transforms.Transform.

 required

VesselMNIST3D

VesselMNIST3D

Bases: MedMNIST

from_batch(batch) classmethod

Instantiate a dataset from a batch generated by a data loader.

Parameters:

Name Type Description Default
batch dict

Dictionary generated by a data loader, containing data that can be converted to instances of [Subject][.torchio.Subject].

 required

dry_iter()

Return the internal list of subjects.

This can be used to iterate over the subjects without loading the data and applying any transforms.

Examples:

>>> names = [subject.name for subject in dataset.dry_iter()]

set_transform(transform)

Set the transform attribute.

Parameters:

Name Type Description Default
transform Callable | None

Callable object, typically an subclass of torchio.transforms.Transform.

 required

SynapseMNIST3D

SynapseMNIST3D

Bases: MedMNIST

from_batch(batch) classmethod

Instantiate a dataset from a batch generated by a data loader.

Parameters:

Name Type Description Default
batch dict

Dictionary generated by a data loader, containing data that can be converted to instances of [Subject][.torchio.Subject].

 required

dry_iter()

Return the internal list of subjects.

This can be used to iterate over the subjects without loading the data and applying any transforms.

Examples:

>>> names = [subject.name for subject in dataset.dry_iter()]

set_transform(transform)

Set the transform attribute.

Parameters:

Name Type Description Default
transform Callable | None

Callable object, typically an subclass of torchio.transforms.Transform.

 required

ZonePlate

ZonePlate

ZonePlate

Bases: Subject

Synthetic data generated from a zone plate.

The zone plate is a circular diffraction grating that produces concentric rings of light and dark bands. This dataset is useful for testing image processing algorithms, particularly those related to frequency analysis and interpolation.

See equation 10.63 in Practical Handbook on Image Processing for Scientific Applications by Bernd Jähne.

Parameters:

Name Type Description Default
size int

The size of the generated image along all dimensions.

 501

shape property

Return shape of first image in subject.

Consistency of shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.shape
(1, 181, 217, 181)

spatial_shape property

Return spatial shape of first image in subject.

Consistency of spatial shapes across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Colin27()
>>> colin.spatial_shape
(181, 217, 181)

spacing property

Return spacing of first image in subject.

Consistency of spacings across images in the subject is checked first.

Examples:

>>> import torchio as tio
>>> colin = tio.datasets.Slicer()
>>> colin.spacing
(1.0, 1.0, 1.2999954223632812)

get_inverse_transform(warn=True, ignore_intensity=False, image_interpolation=None)

Get a reversed list of the inverses of the applied transforms.

Parameters:

Name Type Description Default
warn bool

Issue a warning if some transforms are not invertible.

 True
ignore_intensity bool

If True, all instances of [IntensityTransform][torchio.transforms.intensity_transform.IntensityTransform] will be ignored.

 False
image_interpolation str | None

Modify interpolation for scalar images inside transforms that perform resampling.

 None

apply_inverse_transform(**kwargs)

Apply the inverse of all applied transforms, in reverse order.

Parameters:

Name Type Description Default
**kwargs

Keyword arguments passed on to get_inverse_transform().

 {}

check_consistent_attribute(attribute, relative_tolerance=1e-06, absolute_tolerance=1e-06, message=None)

Check for consistency of an attribute across all images.

Parameters:

Name Type Description Default
attribute str

Name of the image attribute to check

 required
relative_tolerance float

Relative tolerance for numpy.allclose()

 1e-06
absolute_tolerance float

Absolute tolerance for numpy.allclose()

 1e-06

Examples:

>>> import numpy as np
>>> import torch
>>> import torchio as tio
>>> scalars = torch.randn(1, 512, 512, 100)
>>> mask = torch.tensor(scalars > 0).type(torch.int16)
>>> af1 = np.eye([0.8, 0.8, 2.50000000000001, 1])
>>> af2 = np.eye([0.8, 0.8, 2.49999999999999, 1])  # small difference here (e.g. due to different reader)
>>> subject = tio.Subject(
...   image = tio.ScalarImage(tensor=scalars, affine=af1),
...   mask = tio.LabelMap(tensor=mask, affine=af2)
... )
>>> subject.check_consistent_attribute('spacing')  # no error as tolerances are > 0
Note

To check that all values for a specific attribute are close between all images in the subject, numpy.allclose() is used. This function returns True if \(|a_i - b_i| \leq t_{abs} + t_{rel} * |b_i|\), where \(a_i\) and \(b_i\) are the \(i\)-th element of the same attribute of two images being compared, \(t_{abs}\) is the absolute_tolerance and \(t_{rel}\) is the relative_tolerance.

get_image(image_name)

Get a single image by its name.

load()

Load images in subject on RAM.

unload()

Unload images in subject.

add_image(image, image_name)

Add an image to the subject instance.

remove_image(image_name)

Remove an image from the subject instance.

plot(return_fig=False, **kwargs)

Plot images using matplotlib.

Parameters:

Name Type Description Default
return_fig bool

If True, return the figure instead of showing it.

 False
**kwargs

Keyword arguments that will be passed on to plot().

 {}