load_ismrmd#

deepinv.utils.load_ismrmd(fname, data_name='kspace', data_slice=None, **kwargs)[source]#

Load complex MRI data from ISMRMD format.

Uses h5py to load data specified by data_name key. The data is assumed to be stored in complex type.

Note

To speed up loading, slice/index the data before converting to tensor.

Parameters:

fname (str, pathlib.Path) – file to load.
data_name (str) – key of data in file, defaults to “kspace”.
data_slice (tuple) – slice or index to use before converting to tensor, such as int, slice or tuple of these.

Returns:

data loaded in torch.Tensor of shape (2, ...) containing real and imaginary parts, where ... are dimensions of the raw data.

Return type:

Tensor

Examples using `load_ismrmd`:#