Exploring Deep Registration Latent Spaces

  title={Exploring Deep Registration Latent Spaces},
  author={Th{\'e}o Estienne and Maria Vakalopoulou and Stergios Christodoulidis and Enzo Battistella and Th'eophraste Henry and Marvin Lerousseau and Amaury Leroy and Guillaume Chassagnon and Marie-Pierre Revel and Nikos Paragios and Eric Deutsch},
Explainability of deep neural networks is one of the most challenging and interesting problems in the field. In this study, we investigate the topic focusing on the interpretability of deep learning-based registration methods. In particular, with the appropriate model architecture and using a simple linear projection, we decompose the encoding space, generating a new basis, and we empirically show that this basis captures various decomposed anatomically aware geometrical transformations. We… Expand

Figures from this paper


U-ReSNet: Ultimate Coupling of Registration and Segmentation with Deep Nets
The proposed U-ReSNet network learns to automatically generate linear and elastic deformation models, trained by minimizing the mean square error and the local cross correlation similarity metrics, leading to promising results compared to other deep learning-based methods existing in the literature. Expand
Learning a Probabilistic Model for Diffeomorphic Registration
The unsupervised method uses a conditional variational autoencoder network and constrain transformations to be symmetric and diffeomorphic by applying a differentiable exponentiation layer with a symmetric loss function and provides multi-scale velocity field estimations. Expand
Spatial Transformer Networks
This work introduces a new learnable module, the Spatial Transformer, which explicitly allows the spatial manipulation of data within the network, and can be inserted into existing convolutional architectures, giving neural networks the ability to actively spatially transform feature maps. Expand
Deep Learning-Based Concurrent Brain Registration and Tumor Segmentation
A novel, efficient, and multi-task algorithm that addresses the problems of image registration and brain tumor segmentation jointly by exploiting the dependencies between these tasks through a natural coupling of their interdependencies during inference. Expand
Learning Deep Features for Discriminative Localization
In this work, we revisit the global average pooling layer proposed in [13], and shed light on how it explicitly enables the convolutional neural network (CNN) to have remarkable localization abilityExpand
VoxelMorph: A Learning Framework for Deformable Medical Image Registration
VoxelMorph promises to speed up medical image analysis and processing pipelines while facilitating novel directions in learning-based registration and its applications and demonstrates that the unsupervised model’s accuracy is comparable to the state-of-the-art methods while operating orders of magnitude faster. Expand
Linear and Deformable Image Registration with 3D Convolutional Neural Networks
A novel convolutional neural network architecture that couples linear and deformable registration within a unified architecture endowed with near real-time performance is proposed and evaluated on the challenging problem of MRI lung registration. Expand
Unsupervised Deformable Registration for Multi-Modal Images via Disentangled Representations
This paper proposes a fully unsupervised multi-modal deformable image registration method (UMDIR), which does not require any ground truth deformation fields or any aligned multi- modal image pairs during training, and achieves competitive performance against other methods at substantially reduced computation time. Expand
Closing the Gap between Deep and Conventional Image Registration using Probabilistic Dense Displacement Networks
This work proposes to design a network with approximate min-convolutions and mean field inference for differentiable displacement regularisation within a discrete weakly-supervised registration setting that achieves state-of-the-art accuracies for the challenging inter-patient registration of abdominal CT outperforming previous deep learning approaches by 15% Dice overlap. Expand
3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation
The proposed network extends the previous u-net architecture from Ronneberger et al. by replacing all 2D operations with their 3D counterparts and performs on-the-fly elastic deformations for efficient data augmentation during training. Expand