Skip to content

KeepLargestComponent

Bases: Transform

Keep only the largest connected component of each label.

For each specified label value, connected-component analysis is performed and all but the largest component are removed (set to the background value). This is useful for cleaning up noisy segmentation predictions.

Only single-channel LabelMap images are affected.

Parameters:

Name Type Description Default
labels Sequence[int] | None

Label values to filter. None means all non-zero labels found in the data.

 None
background_label int

Value used for removed components.

 0
fully_connected bool

If True, use 26-connectivity (voxels sharing a corner are connected). If False, use 6-connectivity (face-connected only).

 True
**kwargs Any

See Transform.

 {}

Raises:

Type Description
RuntimeError

If a label map has more than one channel.

Examples:

>>> import torchio as tio
>>> transform = tio.KeepLargestComponent()
>>> transform = tio.KeepLargestComponent(labels=[1, 2])
Source code in src/torchio/transforms/label/keep_largest.py 
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
class KeepLargestComponent(Transform):
    r"""Keep only the largest connected component of each label.

    For each specified label value, connected-component analysis is
    performed and all but the largest component are removed (set to
    the background value).  This is useful for cleaning up noisy
    segmentation predictions.

    Only single-channel [`LabelMap`][torchio.LabelMap] images are
    affected.

    Args:
        labels: Label values to filter.  `None` means all non-zero
            labels found in the data.
        background_label: Value used for removed components.
        fully_connected: If `True`, use 26-connectivity (voxels
            sharing a corner are connected).  If `False`, use
            6-connectivity (face-connected only).
        **kwargs: See [`Transform`][torchio.Transform].

    Raises:
        RuntimeError: If a label map has more than one channel.

    Examples:
        >>> import torchio as tio
        >>> transform = tio.KeepLargestComponent()
        >>> transform = tio.KeepLargestComponent(labels=[1, 2])
    """

    def __init__(
        self,
        labels: Sequence[int] | None = None,
        *,
        background_label: int = 0,
        fully_connected: bool = True,
        **kwargs: Any,
    ) -> None:
        super().__init__(**kwargs)
        self.labels = list(labels) if labels is not None else None
        self.background_label = background_label
        self.fully_connected = fully_connected

    def make_params(self, batch: SubjectsBatch) -> dict[str, Any]:
        """No random parameters."""
        return {}

    def apply_transform(
        self,
        batch: SubjectsBatch,
        params: dict[str, Any],
    ) -> SubjectsBatch:
        """Keep only the largest connected component per label."""
        for _name, img_batch in batch.images.items():
            if not issubclass(img_batch._image_class, LabelMap):
                continue
            b, c = img_batch.data.shape[:2]
            if c != 1:
                msg = (
                    "KeepLargestComponent requires single-channel"
                    f" label maps, got {c} channels"
                )
                raise RuntimeError(msg)
            for i in range(b):
                img_batch.data[i, 0] = _keep_largest_per_label(
                    img_batch.data[i, 0],
                    labels=self.labels,
                    background_label=self.background_label,
                    fully_connected=self.fully_connected,
                )
        return batch

supports_per_instance_params property

Whether this transform can sample parameters per batch element.

Defaults to False. Transforms that implement per-instance parameter sampling override this to return True. When False, the transform always uses batch-shared parameters regardless of the per_instance flag, preserving the legacy behavior.

supports_per_instance_p property

Whether this transform can gate each batch element independently.

Defaults to False. Shape-preserving transforms that implement per-element probability override this to return True. Shape-changing transforms must leave it False because masked and unmasked elements would have incompatible shapes.

invertible property

Whether this transform can be inverted.

forward(data) 

forward(data: Subject) -> Subject

forward(data: Image) -> Image

forward(data: Tensor) -> Tensor

forward(data: np.ndarray) -> np.ndarray

forward(data: sitk.Image) -> sitk.Image

forward(data: nib.Nifti1Image) -> nib.Nifti1Image

forward(data: dict) -> dict

forward(data: ImagesBatch) -> ImagesBatch

forward(data: SubjectsBatch) -> SubjectsBatch

Apply the transform.

The output type always matches the input type.

Parameters:

Name Type Description Default
data Any

Input data to transform.

 required
Source code in src/torchio/transforms/transform.py 
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
def forward(self, data: Any) -> Any:
    """Apply the transform.

    The output type always matches the input type.

    Args:
        data: Input data to transform.
    """
    if self.copy:
        data = _copy.deepcopy(data)
    batch, unwrap = self._wrap(data)
    # When per-element gating is active, the transform handles the
    # probability itself (masked-out elements get identity params),
    # so skip the batch-wide coin flip here. Apply iff rand < p, so
    # p=0 is always a no-op and p=1 always applies.
    if not self._per_instance_p_active(batch) and torch.rand(1).item() >= self.p:
        return unwrap(batch)
    params = self.make_params(batch)
    batch = self.apply_transform(batch, params)
    # Record history on the batch, unless every element was gated out by
    # per-element probability: that is an exact no-op, and recording it
    # would let history replay (e.g. an invertible spatial transform)
    # trigger an unnecessary identity resample.
    if not _all_elements_gated_out(params):
        trace = AppliedTransform(name=type(self).__name__, params=params)
        if not hasattr(batch, "applied_transforms"):
            batch.applied_transforms = []
        batch.applied_transforms.append(trace)
    result = unwrap(batch)
    # Propagate history to outputs that can carry it
    if (
        hasattr(batch, "applied_transforms")
        and not isinstance(result, (SubjectsBatch, Tensor, np.ndarray))
        and not isinstance(result, dict)
    ):
        with contextlib.suppress(AttributeError):
            result.applied_transforms = list(batch.applied_transforms)
    return result

inverse(params)

Return a transform that undoes this one.

Override in invertible subclasses. The returned transform, when applied, reverses the effect of the forward pass with the given parameters.

Parameters:

Name Type Description Default
params dict[str, Any]

The parameters recorded in the forward pass.

 required

Returns:

Type Description
Transform

A new Transform instance that inverts this one.
Source code in src/torchio/transforms/transform.py 
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
def inverse(self, params: dict[str, Any]) -> Transform:
    """Return a transform that undoes this one.

    Override in invertible subclasses. The returned transform,
    when applied, reverses the effect of the forward pass with
    the given parameters.

    Args:
        params: The parameters recorded in the forward pass.

    Returns:
        A new `Transform` instance that inverts this one.
    """
    msg = f"{type(self).__name__} is not invertible"
    raise NotImplementedError(msg)

to_hydra()

Export as a Hydra-compatible config dict.

Returns a dict with _target_ set to the fully qualified class name and only non-default field values included.

Returns:

Type Description
dict[str, Any]

Dict suitable for hydra.utils.instantiate().
Source code in src/torchio/transforms/transform.py 
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
def to_hydra(self) -> dict[str, Any]:
    """Export as a Hydra-compatible config dict.

    Returns a dict with `_target_` set to the fully qualified
    class name and only non-default field values included.

    Returns:
        Dict suitable for `hydra.utils.instantiate()`.
    """
    from .parameter_range import _ParameterRange

    cls = type(self)
    target = f"torchio.{cls.__qualname__}"
    cfg: dict[str, Any] = {"_target_": target}

    for name, default in _collect_init_params(cls).items():
        value = getattr(self, name, default)
        if isinstance(value, _ParameterRange):
            if value._original == default:
                continue
            value = _hydra_value(value._original)
        elif value == default:
            continue
        else:
            value = _hydra_value(value)
        cfg[name] = value
    return cfg

make_params(batch)

No random parameters.

Source code in src/torchio/transforms/label/keep_largest.py 
59
60
61
def make_params(self, batch: SubjectsBatch) -> dict[str, Any]:
    """No random parameters."""
    return {}

apply_transform(batch, params)

Keep only the largest connected component per label.

Source code in src/torchio/transforms/label/keep_largest.py 
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
def apply_transform(
    self,
    batch: SubjectsBatch,
    params: dict[str, Any],
) -> SubjectsBatch:
    """Keep only the largest connected component per label."""
    for _name, img_batch in batch.images.items():
        if not issubclass(img_batch._image_class, LabelMap):
            continue
        b, c = img_batch.data.shape[:2]
        if c != 1:
            msg = (
                "KeepLargestComponent requires single-channel"
                f" label maps, got {c} channels"
            )
            raise RuntimeError(msg)
        for i in range(b):
            img_batch.data[i, 0] = _keep_largest_per_label(
                img_batch.data[i, 0],
                labels=self.labels,
                background_label=self.background_label,
                fully_connected=self.fully_connected,
            )
    return batch