mridc.collections.reconstruction.models.convrecnet package

Submodules

mridc.collections.reconstruction.models.convrecnet.crnn_block module

class mridc.collections.reconstruction.models.convrecnet.crnn_block.DataConsistencyLayer[source]

Bases: Module

Data consistency layer for the CRNN. This layer is used to ensure that the output of the CRNN is the same as the input.

forward(pred_kspace, ref_kspace, mask)[source]

Forward pass of the data consistency layer.

training: bool
class mridc.collections.reconstruction.models.convrecnet.crnn_block.RecurrentConvolutionalNetBlock(model: Module, num_iterations: int = 10, fft_centered: bool = True, fft_normalization: str = 'ortho', spatial_dims: Optional[Tuple[int, int]] = None, coil_dim: int = 1, no_dc: bool = False)[source]

Bases: Module

Model block for Recurrent Convolution Neural Network inspired by 1.

References

1

  1. Qin, J. Schlemper, J. Caballero, A. N. Price, J. V. Hajnal and D. Rueckert, “Convolutional Recurrent Neural Networks for Dynamic MR Image Reconstruction,” in IEEE Transactions on Medical Imaging, vol. 38, no. 1, pp. 280-290, Jan. 2019, doi: 10.1109/TMI.2018.2863670.

forward(ref_kspace: Tensor, sens_maps: Tensor, mask: Tensor) → List[Union[Tensor, Any]][source]

Forward pass of the model.

Parameters

  • ref_kspace (Reference k-space data.) –

  • sens_maps (Sensitivity maps.) –

  • mask (Mask to apply to the data.) –

Return type

Reconstructed image.

sens_expand(x: Tensor, sens_maps: Tensor) → Tensor[source]

Expand the sensitivity maps to the same size as the input.

Parameters

  • x (Input data.) –

  • sens_maps (Sensitivity maps.) –

Return type

SENSE reconstruction expanded to the same size as the input.

sens_reduce(x: Tensor, sens_maps: Tensor) → Tensor[source]

Reduce the sensitivity maps to the same size as the input.

Parameters

  • x (Input data.) –

  • sens_maps (Sensitivity maps.) –

Return type

SENSE reconstruction reduced to the same size as the input.

training: bool

Module contents