dataset

dataset

PerfusionDataset container for imaging data and metadata.

This module provides the core data container for perfusion imaging data, supporting 3D and 4D arrays with associated acquisition metadata.

PerfusionDataset dataclass

PerfusionDataset(
    data,
    affine,
    modality,
    time_points=None,
    acquisition_params=AcquisitionParams(),
    source_path=None,
    source_format="unknown",
    quality_mask=None,
)

Container for perfusion imaging data and metadata.

This dataclass holds the loaded imaging data along with spatial transformation, temporal information, and acquisition parameters.

ATTRIBUTE	DESCRIPTION
data	3D or 4D array of image data with shape (x, y, z) or (x, y, z, t). TYPE: NDArray[floating]
affine	4x4 affine transformation matrix mapping voxel to world coordinates. TYPE: NDArray[floating]
modality	Perfusion imaging modality (DCE, DSC, ASL, IVIM). TYPE: Modality
time_points	Time vector in seconds for dynamic (4D) data. Must have length equal to data.shape[3] if data is 4D. TYPE: NDArray[floating] | None
acquisition_params	Modality-specific acquisition parameters. TYPE: AnyAcquisitionParams
source_path	Original file path if loaded from disk. TYPE: Path | None
source_format	Source file format ('nifti', 'dicom', 'bids'). TYPE: str
quality_mask	Boolean mask of valid voxels. Shape must match spatial dimensions. TYPE: NDArray[bool_] | None

RAISES	DESCRIPTION
ValueError	If data is not 3D or 4D, if affine is not 4x4, or if time_points length does not match 4th dimension.

Examples:

>>> import numpy as np
>>> from osipy.common.dataset import PerfusionDataset
>>> from osipy.common.types import Modality, DCEAcquisitionParams
>>> data = np.random.rand(64, 64, 20, 30)
>>> affine = np.eye(4)
>>> time = np.linspace(0, 300, 30)
>>> params = DCEAcquisitionParams(tr=5.0, te=2.0, flip_angles=[2, 5, 10, 15])
>>> dataset = PerfusionDataset(
...     data=data,
...     affine=affine,
...     modality=Modality.DCE,
...     time_points=time,
...     acquisition_params=params,
... )

shape property

shape

Return the shape of the data array.

spatial_shape property

spatial_shape

Return the spatial dimensions (x, y, z).

n_timepoints property

n_timepoints

Return the number of time points (0 for 3D data).

is_dynamic property

is_dynamic

Return True if data is 4D (dynamic/time-series).

voxel_size property

voxel_size

Return voxel dimensions in mm from affine matrix.

__post_init__

__post_init__()

Validate dataset after initialization.