torchio.transforms.transform.Transform.parse_sample() - Code Metrics - Inspection of "Add tests and fix bugs in Image (#244)" - fepegar/torchio - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( c291a8...879ee9 )

by Fernando

created 2020-07-31 14:48 UTC

Transform.parse_sample() A

↳ Parent: torchio.transforms.transform

Complexity

Conditions

Size

Total Lines	9
Code Lines	6

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	2
eloc	6
nop	1
dl	0
loc	9
rs	10
c	0
b	0
f	0

import numbers
import warnings
from typing import Union, Tuple
from copy import deepcopy
from abc import ABC, abstractmethod

import torch
import numpy as np
import SimpleITK as sitk

from .. import TypeData, INTENSITY, DATA, TypeNumber
from ..data.image import Image
from ..data.subject import Subject
from ..data.dataset import ImagesDataset
from ..utils import nib_to_sitk, sitk_to_nib
from .interpolation import Interpolation


class Transform(ABC):
    """Abstract class for all TorchIO transforms.

    All classes used to transform a sample from an
    :py:class:`~torchio.ImagesDataset` should subclass it.
    All subclasses should overwrite
    :py:meth:`torchio.tranforms.Transform.apply_transform`,
    which takes a sample, applies some transformation and returns the result.

    Args:
        p: Probability that this transform will be applied.
        copy: Make a deep copy of the input before applying the transform.
    """
    def __init__(self, p: float = 1, copy: bool = True):
        self.probability = self.parse_probability(p)
        self.copy = copy

    def __call__(self, data: Union[Subject, torch.Tensor, np.ndarray]):
        """Transform a sample and return the result.

        Args:
            data: Instance of :py:class:`~torchio.Subject`, 4D
                :py:class:`torch.Tensor` or 4D NumPy array with dimensions
                :math:`(C, D, H, W)`, where :math:`C` is the number of channels
                and :math:`D, H, W` are the spatial dimensions. If the input is
                a tensor, the affine matrix is an identity and a tensor will be
                also returned.
        """
        if torch.rand(1).item() > self.probability:
            return data
        if isinstance(data, (np.ndarray, torch.Tensor)):
            is_array = isinstance(data, np.ndarray)
            is_tensor = True
            sample = self.parse_tensor(data)
        else:
            is_tensor = is_array = False
            sample = data
        self.parse_sample(sample)

        if self.copy:
            sample = deepcopy(sample)

        with np.errstate(all='raise'):
            transformed = self.apply_transform(sample)

        if is_tensor:
            num_channels = len(data)
            images = [
                transformed[f'channel_{i}'][DATA]
                for i in range(num_channels)
            ]
            transformed = torch.cat(images)
        if is_array:
            transformed = transformed.numpy()
        return transformed

    @abstractmethod
    def apply_transform(self, sample: Subject):
        raise NotImplementedError

    @staticmethod
    def parse_range(
            nums_range: Union[TypeNumber, Tuple[TypeNumber, TypeNumber]],
            name: str,
            min_constraint: TypeNumber = None,
            max_constraint: TypeNumber = None,
            type_constraint: type = None,
            ) -> Tuple[TypeNumber, TypeNumber]:
        r"""Adapted from ``torchvision.transforms.RandomRotation``.

        Args:
            nums_range: Tuple of two numbers :math:`(n_{min}, n_{max})`,
                where :math:`n_{min} \leq n_{max}`.
                If a single positive number :math:`n` is provided,
                :math:`n_{min} = -n` and :math:`n_{max} = n`.
            name: Name of the parameter, so that an informative error message
                can be printed.
            min_constraint: Minimal value that :math:`n_{min}` can take,
                default is None, i.e. there is no minimal value.
            max_constraint: Maximal value that :math:`n_{max}` can take,
                default is None, i.e. there is no maximal value.
            type_constraint: Precise type that :math:`n_{max}` and
                :math:`n_{min}` must take.

        Returns:
            A tuple of two numbers :math:`(n_{min}, n_{max})`.

        Raises:
            ValueError: if :attr:`nums_range` is negative
            ValueError: if :math:`n_{max}` or :math:`n_{min}` is not a number
            ValueError: if :math:`n_{max} \lt n_{min}`
            ValueError: if :attr:`min_constraint` is not None and
                :math:`n_{min}` is smaller than :attr:`min_constraint`
            ValueError: if :attr:`max_constraint` is not None and
                :math:`n_{max}` is greater than :attr:`max_constraint`
            ValueError: if :attr:`type_constraint` is not None and
                :math:`n_{max}` and :math:`n_{max}` are not of type
                :attr:`type_constraint`.
        """
        if isinstance(nums_range, numbers.Number):
            if nums_range < 0:
                raise ValueError(
                    f'If {name} is a single number,'
                    f' it must be positive, not {nums_range}')
            if min_constraint is not None and nums_range < min_constraint:
                raise ValueError(
                    f'If {name} is a single number, it must be greater'
                    f'than {min_constraint}, not {nums_range}'
                )
            if max_constraint is not None and nums_range > max_constraint:
                raise ValueError(
                    f'If {name} is a single number, it must be smaller'
                    f'than {max_constraint}, not {nums_range}'
                )
            if type_constraint is not None and \
                    not isinstance(nums_range, type_constraint):
                raise ValueError(
                    f'If {name} is a single number, it must be of'
                    f'type {type_constraint}, not {nums_range}'
                )
            min_range = -nums_range if min_constraint is None else nums_range
            return (min_range, nums_range)

        try:
            min_degree, max_degree = nums_range
        except (TypeError, ValueError):
            raise ValueError(
                f'If {name} is not a single number, it muste be'
                f'a sequence of len 2, not {nums_range}'
            )

        if not isinstance(min_degree, numbers.Number) or \
                not isinstance(max_degree, numbers.Number):
            message = (
                f'{name} values must be numbers, not {nums_range}')
            raise ValueError(message)

        if min_degree > max_degree:
            raise ValueError(
                f'If {name} is a sequence, the second value must be'
                f' equal or greater than the first, not {nums_range}')

        if min_constraint is not None and min_degree < min_constraint:
            raise ValueError(
                f'If {name} is a sequence, the first value must be greater'
                f'than {min_constraint}, not {min_degree}'
            )

        if max_constraint is not None and max_degree > max_constraint:
            raise ValueError(
                f'If {name} is a sequence, the second value must be smaller'
                f'than {max_constraint}, not {max_degree}'
            )

        if type_constraint is not None:
            if not isinstance(min_degree, type_constraint) or \
                    not isinstance(max_degree, type_constraint):
                raise ValueError(
                    f'If {name} is a sequence, its values must be of'
                    f'type {type_constraint}, not {nums_range}'
                )
        return nums_range

    @staticmethod
    def parse_probability(probability: float) -> float:
        is_number = isinstance(probability, numbers.Number)
        if not (is_number and 0 <= probability <= 1):
            message = (
                'Probability must be a number in [0, 1],'
                f' not {probability}'
            )
            raise ValueError(message)
        return probability

    @staticmethod
    def parse_sample(sample: Subject) -> None:
        if not isinstance(sample, Subject):
            message = (
                'Input to a transform must be a PyTorch tensor or an instance'
                ' of torchio.Subject generated by a torchio.ImagesDataset,'
                f' not "{type(sample)}"'
            )
            raise RuntimeError(message)

    def parse_tensor(self, data: TypeData) -> Subject:
        if isinstance(data, np.ndarray):
            tensor = torch.from_numpy(data)
        else:
            tensor = data
        tensor = tensor.float()  # does nothing if already float
        num_dimensions = tensor.dim()
        if num_dimensions != 4:
            message = (
                'The input tensor must have 4 dimensions (channels, i, j, k),'
                f' but has {num_dimensions}: {tensor.shape}'
            )
            raise RuntimeError(message)
        return self._get_subject_from_tensor(tensor)

    @staticmethod
    def parse_interpolation(interpolation: str) -> Interpolation:
        if isinstance(interpolation, Interpolation):
            message = (
                'Interpolation of type torchio.Interpolation'
                ' is deprecated, please use a string instead'
            )
            warnings.warn(message, FutureWarning)
        elif isinstance(interpolation, str):
            interpolation = interpolation.lower()
            supported_values = [key.name.lower() for key in Interpolation]
            if interpolation in supported_values:
                interpolation = getattr(Interpolation, interpolation.upper())
            else:
                message = (
                    f'Interpolation "{interpolation}" is not among'
                    f' the supported values: {supported_values}'
                )
                raise AttributeError(message)
        else:
            message = (
                'image_interpolation must be a string,'
                f' not {type(interpolation)}'
            )
            raise TypeError(message)
        return interpolation

    @staticmethod
    def _get_subject_from_tensor(tensor: torch.Tensor) -> Subject:
        subject_dict = {}
        for channel_index, channel_tensor in enumerate(tensor):
            name = f'channel_{channel_index}'
            image = Image(tensor=channel_tensor, type=INTENSITY)
            subject_dict[name] = image
        subject = Subject(subject_dict)
        return subject

    @staticmethod
    def nib_to_sitk(data: TypeData, affine: TypeData):
        return nib_to_sitk(data, affine)

    @staticmethod
    def sitk_to_nib(image: sitk.Image):
        return sitk_to_nib(image)

    @property
    def name(self):
        return self.__class__.__name__


1			import numbers
2			import warnings
3			from typing import Union, Tuple
4			from copy import deepcopy
5			from abc import ABC, abstractmethod
6
7			import torch
8			import numpy as np
9			import SimpleITK as sitk
10
11			from .. import TypeData, INTENSITY, DATA, TypeNumber
12			from ..data.image import Image
13			from ..data.subject import Subject
14			from ..data.dataset import ImagesDataset
15			from ..utils import nib_to_sitk, sitk_to_nib
16			from .interpolation import Interpolation
17
18
19			class Transform(ABC):
20			"""Abstract class for all TorchIO transforms.
21
22			All classes used to transform a sample from an
23			:py:class:`~torchio.ImagesDataset` should subclass it.
24			All subclasses should overwrite
25			:py:meth:`torchio.tranforms.Transform.apply_transform`,
26			which takes a sample, applies some transformation and returns the result.
27
28			Args:
29			p: Probability that this transform will be applied.
30			copy: Make a deep copy of the input before applying the transform.
31			"""
32			def __init__(self, p: float = 1, copy: bool = True):
33			self.probability = self.parse_probability(p)
34			self.copy = copy
35
36			def __call__(self, data: Union[Subject, torch.Tensor, np.ndarray]):
37			"""Transform a sample and return the result.
38
39			Args:
40			data: Instance of :py:class:`~torchio.Subject`, 4D
41			:py:class:`torch.Tensor` or 4D NumPy array with dimensions
42			:math:`(C, D, H, W)`, where :math:`C` is the number of channels
43			and :math:`D, H, W` are the spatial dimensions. If the input is
44			a tensor, the affine matrix is an identity and a tensor will be
45			also returned.
46			"""
47			if torch.rand(1).item() > self.probability:
48			return data
49			if isinstance(data, (np.ndarray, torch.Tensor)):
50			is_array = isinstance(data, np.ndarray)
51			is_tensor = True
52			sample = self.parse_tensor(data)
53			else:
54			is_tensor = is_array = False
55			sample = data
56			self.parse_sample(sample)
57
58			if self.copy:
59			sample = deepcopy(sample)
60
61			with np.errstate(all='raise'):
62			transformed = self.apply_transform(sample)
63
64			if is_tensor:
65			num_channels = len(data)
66			images = [
67			transformed[f'channel_{i}'][DATA]
68			for i in range(num_channels)
69			]
70			transformed = torch.cat(images)
71			if is_array:
72			transformed = transformed.numpy()
73			return transformed
74
75			@abstractmethod
76			def apply_transform(self, sample: Subject):
77			raise NotImplementedError
78
79			@staticmethod
80			def parse_range(
81			nums_range: Union[TypeNumber, Tuple[TypeNumber, TypeNumber]],
82			name: str,
83			min_constraint: TypeNumber = None,
84			max_constraint: TypeNumber = None,
85			type_constraint: type = None,
86			) -> Tuple[TypeNumber, TypeNumber]:
87			r"""Adapted from ``torchvision.transforms.RandomRotation``.
88
89			Args:
90			nums_range: Tuple of two numbers :math:`(n_{min}, n_{max})`,
91			where :math:`n_{min} \leq n_{max}`.
92			If a single positive number :math:`n` is provided,
93			:math:`n_{min} = -n` and :math:`n_{max} = n`.
94			name: Name of the parameter, so that an informative error message
95			can be printed.
96			min_constraint: Minimal value that :math:`n_{min}` can take,
97			default is None, i.e. there is no minimal value.
98			max_constraint: Maximal value that :math:`n_{max}` can take,
99			default is None, i.e. there is no maximal value.
100			type_constraint: Precise type that :math:`n_{max}` and
101			:math:`n_{min}` must take.
102
103			Returns:
104			A tuple of two numbers :math:`(n_{min}, n_{max})`.
105
106			Raises:
107			ValueError: if :attr:`nums_range` is negative
108			ValueError: if :math:`n_{max}` or :math:`n_{min}` is not a number
109			ValueError: if :math:`n_{max} \lt n_{min}`
110			ValueError: if :attr:`min_constraint` is not None and
111			:math:`n_{min}` is smaller than :attr:`min_constraint`
112			ValueError: if :attr:`max_constraint` is not None and
113			:math:`n_{max}` is greater than :attr:`max_constraint`
114			ValueError: if :attr:`type_constraint` is not None and
115			:math:`n_{max}` and :math:`n_{max}` are not of type
116			:attr:`type_constraint`.
117			"""
118			if isinstance(nums_range, numbers.Number):
119			if nums_range < 0:
120			raise ValueError(
121			f'If {name} is a single number,'
122			f' it must be positive, not {nums_range}')
123			if min_constraint is not None and nums_range < min_constraint:
124			raise ValueError(
125			f'If {name} is a single number, it must be greater'
126			f'than {min_constraint}, not {nums_range}'
127			)
128			if max_constraint is not None and nums_range > max_constraint:
129			raise ValueError(
130			f'If {name} is a single number, it must be smaller'
131			f'than {max_constraint}, not {nums_range}'
132			)
133			if type_constraint is not None and \
134			not isinstance(nums_range, type_constraint):
135			raise ValueError(
136			f'If {name} is a single number, it must be of'
137			f'type {type_constraint}, not {nums_range}'
138			)
139			min_range = -nums_range if min_constraint is None else nums_range
140			return (min_range, nums_range)
141
142			try:
143			min_degree, max_degree = nums_range
144			except (TypeError, ValueError):
145			raise ValueError(
146			f'If {name} is not a single number, it muste be'
147			f'a sequence of len 2, not {nums_range}'
148			)
149
150			if not isinstance(min_degree, numbers.Number) or \
151			not isinstance(max_degree, numbers.Number):
152			message = (
153			f'{name} values must be numbers, not {nums_range}')
154			raise ValueError(message)
155
156			if min_degree > max_degree:
157			raise ValueError(
158			f'If {name} is a sequence, the second value must be'
159			f' equal or greater than the first, not {nums_range}')
160
161			if min_constraint is not None and min_degree < min_constraint:
162			raise ValueError(
163			f'If {name} is a sequence, the first value must be greater'
164			f'than {min_constraint}, not {min_degree}'
165			)
166
167			if max_constraint is not None and max_degree > max_constraint:
168			raise ValueError(
169			f'If {name} is a sequence, the second value must be smaller'
170			f'than {max_constraint}, not {max_degree}'
171			)
172
173			if type_constraint is not None:
174			if not isinstance(min_degree, type_constraint) or \
175			not isinstance(max_degree, type_constraint):
176			raise ValueError(
177			f'If {name} is a sequence, its values must be of'
178			f'type {type_constraint}, not {nums_range}'
179			)
180			return nums_range
181
182			@staticmethod
183			def parse_probability(probability: float) -> float:
184			is_number = isinstance(probability, numbers.Number)
185			if not (is_number and 0 <= probability <= 1):
186			message = (
187			'Probability must be a number in [0, 1],'
188			f' not {probability}'
189			)
190			raise ValueError(message)
191			return probability
192
193			@staticmethod
194			def parse_sample(sample: Subject) -> None:
195			if not isinstance(sample, Subject):
196			message = (
197			'Input to a transform must be a PyTorch tensor or an instance'
198			' of torchio.Subject generated by a torchio.ImagesDataset,'
199			f' not "{type(sample)}"'
200			)
201			raise RuntimeError(message)
202
203			def parse_tensor(self, data: TypeData) -> Subject:
204			if isinstance(data, np.ndarray):
205			tensor = torch.from_numpy(data)
206			else:
207			tensor = data
208			tensor = tensor.float() # does nothing if already float
209			num_dimensions = tensor.dim()
210			if num_dimensions != 4:
211			message = (
212			'The input tensor must have 4 dimensions (channels, i, j, k),'
213			f' but has {num_dimensions}: {tensor.shape}'
214			)
215			raise RuntimeError(message)
216			return self._get_subject_from_tensor(tensor)
217
218			@staticmethod
219			def parse_interpolation(interpolation: str) -> Interpolation:
220			if isinstance(interpolation, Interpolation):
221			message = (
222			'Interpolation of type torchio.Interpolation'
223			' is deprecated, please use a string instead'
224			)
225			warnings.warn(message, FutureWarning)
226			elif isinstance(interpolation, str):
227			interpolation = interpolation.lower()
228			supported_values = [key.name.lower() for key in Interpolation]
229			if interpolation in supported_values:
230			interpolation = getattr(Interpolation, interpolation.upper())
231			else:
232			message = (
233			f'Interpolation "{interpolation}" is not among'
234			f' the supported values: {supported_values}'
235			)
236			raise AttributeError(message)
237			else:
238			message = (
239			'image_interpolation must be a string,'
240			f' not {type(interpolation)}'
241			)
242			raise TypeError(message)
243			return interpolation
244
245			@staticmethod
246			def _get_subject_from_tensor(tensor: torch.Tensor) -> Subject:
247			subject_dict = {}
248			for channel_index, channel_tensor in enumerate(tensor):
249			name = f'channel_{channel_index}'
250			image = Image(tensor=channel_tensor, type=INTENSITY)
251			subject_dict[name] = image
252			subject = Subject(subject_dict)
253			return subject
254
255			@staticmethod
256			def nib_to_sitk(data: TypeData, affine: TypeData):
257			return nib_to_sitk(data, affine)
258
259			@staticmethod
260			def sitk_to_nib(image: sitk.Image):
261			return sitk_to_nib(image)
262
263			@property
264			def name(self):
265			return self.__class__.__name__
266

fepegar / torchio

Push — master ( c291a8...879ee9 )

Transform.parse_sample() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like