MoDL

class deepinv.models.MoDL(denoiser=None, num_iter=3)

Bases: BaseUnfold

MoDL unfolded network.

The model is a simple unrolled network using half-quadratic splitting where the prox is replaced by a trainable denoising prior.

This was proposed for MRI reconstruction in Aggarwal et al.[1].

Parameters:

• denoiser (Denoiser, torch.nn.Module) – backbone denoiser model. If None, uses deepinv.models.DnCNN

• num_iter (int) – number of unfolded layers (“cascades”), defaults to 3.

---

References:

[1]

Hemant K Aggarwal, Merry P Mani, and Mathews Jacob. Modl: model-based deep learning architecture for inverse problems. IEEE transactions on medical imaging, 38(2):394–405, 2018.

Examples using MoDL:

Tour of MRI functionality in DeepInverse

Tour of MRI functionality in DeepInverse

Self-supervised MRI reconstruction with Artifact2Artifact

Self-supervised MRI reconstruction with Artifact2Artifact

Self-supervised learning with Equivariant Imaging for MRI.

Self-supervised learning with Equivariant Imaging for MRI.