Semseg patch fast

This dataloader is the default dataloader, it is inspired from nnUnet but simplified.

Dataset primitives for 3D segmentation dataset. Solution: patch approach with the whole dataset into memory.

class biom3d.datasets.semseg_patch_fast.LabelToLong(label_name: str)[source]

Transform to convert label data to long (integer) type.

Variables:

label_name (str) – Name of the label to be transformed.

__init__(label_name: str)[source]

Transform to convert label data to long (integer) type.

Parameters:

label_name (str) – Name of the label to be transformed.

Returns:

subject – Dictionary with the label data converted to long (integer) type.

Return type:

dict

class biom3d.datasets.semseg_patch_fast.SemSeg3DPatchFast(*args: Any, **kwargs: Any)[source]

Dataset class for semantic segmentation with 3D patches. Supports data augmentation and efficient loading.

Variables:

  • img_path (str) – Path to collection containing the image files.

  • msk_path (str) – Path to collection containing the mask files.

  • fg_path (str | None) – Path to collection containing the foreground files.

  • batch_size (int) – Size of a batch.

  • patch_size (numpy.ndarray) – Size of a patch.

  • aug_patch_size (numpy.ndarray | None) – Size of augmented patch size, may be bigger than patch size.

  • nbof_steps (int) – Number of steps (batches) per epoch.

  • load_data (bool) – If True, load the entire dataset into memory.

  • handler (DataHandler) – DataHandler used to load data.

  • train (bool) – If True, use the dataset for training; otherwise, use it for validation.

  • fnames (list[str]) – List of image paths relative to img_path.

  • use_aug (bool) – Whether to use data augmentation

  • fg_rate (float) – Foreground rate, used to force foreground inclusion in patches.

  • crop_scale (float) – Scale factor for crop size during augmentation.

  • use_softmax (bool) – If True, use softmax activation.

  • batch_idx (int) – Current batch index.

__init__(img_path: str, msk_path: str, batch_size: int, patch_size: ndarray, nbof_steps: int, folds_csv: str | None = None, fold: int = 0, val_split: float = 0.25, train: bool = True, use_aug: bool = True, aug_patch_size: ndarray | None = None, fg_path: str | None = None, fg_rate: float = 0.33, crop_scale: float = 1.0, load_data: bool = False, use_softmax: bool = True)[source]

Dataset class for semantic segmentation with 3D patches. Supports data augmentation and efficient loading.

Parameters:

  • img_path (str) – Path to collection containing the image files.

  • msk_path (str) – Path to collection containing the mask files.

  • batch_size (int) – Batch size for dataset sampling.

  • patch_size (numpy.ndarray) – Size of the patches to be used.

  • nbof_steps (int) – Number of steps (batches) per epoch.

  • folds_csv (str, optional) – CSV file containing fold information for dataset splitting.

  • fold (int, default=0) – The current fold number for training/validation splitting.

  • val_split (float, default=0.25) – Proportion of data to be used for validation.

  • train (bool, default=True) – If True, use the dataset for training; otherwise, use it for validation.

  • use_aug (bool, default=True) – If True, apply data augmentation.

  • aug_patch_size (numpy.ndarray, optional) – Patch size to use for augmented patches.

  • fg_path (str, optional) – Path to collection containing foreground information.

  • fg_rate (float, default=0.33) – Foreground rate, used to force foreground inclusion in patches. If > 0, force the use of foreground, needs to run some pre-computations (note: better use the foreground scheduler)

  • crop_scale (float, default=1.0) – Scale factor for crop size during augmentation. If > 1, then use random_crop_resize instead of random_crop_pad

  • load_data (bool, default=False) – If True, load the entire dataset into memory.

  • use_softmax (bool, default=True) – If True, use softmax activation.

Raises:

AssertionError – If fold_csv is None and not valid path for datas, or empty collections If crop_scale < 1

set_fg_rate(value: float) → None[source]

Setter function for the foreground rate class parameter.

biom3d.datasets.semseg_patch_fast.centered_crop(img: ndarray, msk: ndarray, center: Iterable[int], crop_shape: Iterable[int], margin: Iterable[float] = array([0., 0., 0.])) → tuple[ndarray, ndarray][source]

Do a crop, forcing the location voxel to be located in the center of the crop.

Parameters:

  • img (numpy.ndarray) – Image data.

  • msk (ndaarray) – Mask data.

  • center (iterable of int) – Center voxel location for cropping.

  • crop_shape (iterable of int) – Shape of the crop.

  • margin (iterable of float, default np.zeros(3)) – Margin around the center location.

Raises:

AssertionError – If center is out of the image

Returns:

  • crop_img (numpy.ndarray) – Cropped image data.

  • crop_msk (numpy.ndarray) – Cropped mask data.

biom3d.datasets.semseg_patch_fast.centered_pad(img: ndarray, final_size: Iterable[int], msk: ndarray | None = None) → ndarray | tuple[ndarray, ndarray][source]

Do a centered pad an img and msk to fit the final_size.

Parameters:

  • img (numpy.ndarray) – Image data.

  • final_size (iterable of int) – Final size of the cropped image and mask.

  • msk (numpy.ndarray, optional) – Mask data.

Returns:

  • pad_img (numpy.ndarray) – Cropped image data, focused on the foreground region.

  • pad_msk (numpy.ndarray, optional) – Cropped mask data, corresponding to the cropped image region.

biom3d.datasets.semseg_patch_fast.foreground_crop(img: ndarray, msk: ndarray, final_size: Iterable[int], fg_margin: Iterable[float], fg: dict[int, ndarray] | None = None, use_softmax: bool = True) → tuple[ndarray, ndarray][source]

Do a foreground crop.

Parameters:

  • img (numpy.ndarray) – Image data.

  • msk (numpy.ndarray) – Mask data.

  • final_size (iterable of int) – Final size of the cropped image and mask.

  • fg_margin (iterable of float) – Margin around the foreground location.

  • fg (dict of int to numpy.ndarray, optional) – Foreground information.

  • use_softmax (bool, default=True) – If True, assumes softmax activation.

Returns:

  • img (numpy.ndarray) – Cropped image data, focused on the foreground region.

  • msk (numpy.ndarray) – Cropped mask data, corresponding to the cropped image region.

biom3d.datasets.semseg_patch_fast.located_crop(img: ndarray, msk: ndarray, location: Iterable[int], crop_shape: Iterable[int], margin: Iterable[float] = array([0., 0., 0.])) → tuple[ndarray, ndarray][source]

Do a crop, forcing the location voxel to be located in the crop.

Parameters:

  • img (numpy.ndarray) – Image data.

  • msk (numpy.ndarray) – Mask data.

  • location (iterable of int) – Specific voxel location to include in the crop.

  • crop_shape (iterable of int) – Shape of the crop.

  • margin (iterable of float) – Margin around the location.

Returns:

  • crop_img (numpy.ndarray) – Cropped image data.

  • crop_msk (numpy.ndarray) – Cropped mask data.

biom3d.datasets.semseg_patch_fast.random_crop(img: ndarray, msk: ndarray, crop_shape: Iterable[int], force_in: bool = True) → tuple[ndarray, ndarray][source]

Randomly crop a portion of size prop of the original image size.

Parameters:

  • img (numpy.ndarray) – Image data.

  • msk (numpy.ndarray) – Mask data.

  • crop_shape – Shape of the crop.

  • force_in (bool, optional) – If True, ensures the crop is fully within the image boundaries.

Raises:

AssertionError – If image shape (minus C) is not the same shape as crop_shape

Returns:

  • crop_img (numpy.ndarray) – Cropped image data.

  • crop_msk (numpy.ndarray) – Cropped mask data.

biom3d.datasets.semseg_patch_fast.random_crop_pad(img: ndarray, msk: ndarray, final_size: Iterable[int], fg_rate: float = 0.33, fg_margin: Iterable[float] = array([0., 0., 0.]), fg: dict[int, ndarray] | None = None, use_softmax: bool = True) → tuple[ndarray, ndarray][source]

Do a random crop and pad if needed.

Parameters:

  • img (numpy.ndarray) – Image data.

  • msk (numpy.ndarray) – Mask data.

  • final_size – Final size of the image and mask after cropping and padding.

  • fg_rate (float, default=0.33) – Probability of focusing the crop on the foreground.

  • fg_margin (iterable of float, default=np.zeros(3)) – Margin around the foreground location.

  • fg (dict of int to numpy.ndarray, optional) – Foreground information.

  • use_softmax (bool, default=True) – If True, assumes softmax activation.

Returns:

  • img (numpy.ndarray) – Cropped and padded image data.

  • msk (numpy.ndarray) – Cropped and padded mask data.

biom3d.datasets.semseg_patch_fast.random_crop_resize(img: ndarray, msk: ndarray, crop_scale: float, final_size: Iterable[int], fg_rate: int = 0.33, fg_margin: Iterable[float] = array([0., 0., 0.])) → tuple[ndarray, ndarray][source]

Do a random crop and resize if needed.

Parameters:

  • img (numpy.ndarray) – Image data.

  • msk (numpy.ndarray) – Mask data.

  • crop_scale (float, >=1) – Scale factor for the crop size.

  • final_size (iterable of int) – Final size of the image and mask after cropping and resizing.

  • fg_rate (float, default=0.33) – Probability of focusing the crop on the foreground.

  • fg_margin (iterable of float, default=np.zeros(3)) – Margin around the foreground location.

Raises:

ValueError – If crop_scale < 1.

Returns:

  • img (numpy.ndarray) – Cropped and resized image data.

  • msk (numpy.ndarray) – Cropped and resized mask data.