MIR.models.HyperTransMorph

Hyper-TransMorph model Chen, J., Du, Y., He, Y., Segars, W. P., Li, Y., & Frey, E. C. (2021). TransMorph: Transformer for unsupervised medical image registration. arXiv preprint arXiv:2111.10480. Swin-Transformer code retrieved from: https://github.com/SwinTransformer/Swin-Transformer-Semantic-Segmentation Original paper: Liu, Z., Lin, Y., Cao, Y., Hu, H., Wei, Y., Zhang, Z., … & Guo, B. (2021). Swin transformer: Hierarchical vision transformer using shifted windows. arXiv preprint arXiv:2103.14030. Modified and tested by: Junyu Chen jchen245@jhmi.edu Johns Hopkins University

Functions

window_partition(x, window_size)
window_reverse(windows, window_size, H, W, L)

Classes

BasicLayer(*args, **kwargs)	A basic Swin Transformer layer for one stage.
Conv3dReLU(*args, **kwargs)
CustomConv(*args, **kwargs)	Hyper-conditioned convolution with weights predicted from parameters.
DecoderBlock(*args, **kwargs)
HyperBlocks(*args, **kwargs)	MLP that embeds hyperparameters for hypernetwork conditioning.
HyperLinear(*args, **kwargs)	Linear layer with weights predicted from hyperparameters.
HyperTransMorphTVF(*args, **kwargs)
HyperTransMorphTVFSPR(*args, **kwargs)	TransMorph TVF with Spatially-varying regularization :param config: Configuration object containing model parameters :param time_steps: Number of time steps for progressive registration :param SVF: Boolean indicating whether to use SVF (Time Stationary Velocity Field) integration :param SVF_steps: Number of steps for SVF integration :param composition: Type of composition for flow integration ('composition' or 'addition') :param swin_type: Type of Swin Transformer to use ('swin' or 'dswin')
Mlp(*args, **kwargs)	Feed-forward MLP block used inside transformer blocks.
PatchEmbed(*args, **kwargs)	Image to Patch Embedding :param patch_size: Patch token size.
PatchMerging(*args, **kwargs)	Patch Merging Layer.
RegistrationHead(*args, **kwargs)
SinPositionalEncoding3D(*args, **kwargs)
SinusoidalPositionEmbedding(*args, **kwargs)	Rotary Position Embedding
SwinTransformer(*args, **kwargs)	Swin Transformer
SwinTransformerBlock(*args, **kwargs)	Swin Transformer Block.
WindowAttention(*args, **kwargs)	Window based multi-head self attention (W-MSA) module with relative position bias.