viqa.utils.load_data

viqa.utils.load_data(img: ndarray | ImageArray | Tensor | str | PathLike, data_range: int | None = None, normalize: bool = False, batch: bool = False, roi: list[Tuple[int, int]] | None = None) → list[ImageArray] | ImageArray[source]

Load data from a numpy array, a pytorch tensor or a file path.

Parameters:

  • img (np.ndarray, viqa.ImageArray, torch.Tensor, str or os.PathLike) – Numpy array, ImageArray, tensor or file path

  • data_range (int, optional, default=None) – Maximum value of the returned data. Passed to viqa.utils.normalize_data().

  • normalize (bool, default False) – If True, data is normalized to (0, data_range) based on min and max of img.

  • batch (bool, default False) –

    If True, img is a file path and all files in the directory are loaded.

    Deprecated since version 4.0.0: This will be deprecated in version 4.0.0.

  • roi (list[Tuple[int, int]], optional, default=None) – Region of interest for cropping the image. The format is a list of tuples with the ranges for the x, y (and z) axis. First value in the tuple denotes the start and the second value the end of the range. If not set, the whole image is loaded.

Returns:

img_arr – viqa.utils.ImageArray or list of viqa.utils.ImageArray containing the data

Return type:

ImageArray or list[ImageArray]

Raises:

ValueError – If input type is not supported If data_range=None and normalize=True

Warns:

RuntimeWarning – If data_range is set but normalize=False. data_range will be ignored.

Warning

batch will be deprecated in version 4.0.0.

Examples

>>> from viqa import load_data
>>> path_r = "path/to/reference/image.mhd"
>>> path_m = "path/to/modified/image.mhd"
>>> img_r = load_data(path_r)
>>> img_m = load_data(path_m)

>>> from viqa import load_data
>>> img_r = np.random.rand(128, 128)
>>> img_r = load_data(img_r, data_range=255, normalize=True)