A deep learning framework for unsupervised affine and deformable image registration

Image registration, the process of aligning two or more images, is the core technique of many (semi-)automatic medical image analysis tasks. Recent studies have shown that deep learning methods, notably convolutional neural networks (ConvNets), can be used for image registration. Thus far training of ConvNets for registration was supervised using predefined example registrations. However, obtaining example registrations is not trivial. To circumvent the need for predefined examples, and thereby to increase convenience of training ConvNets for image registration, we propose the Deep Learning Image Registration (DLIR) framework for unsupervised affine and deformable image registration. In the DLIR framework ConvNets are trained for image registration by exploiting image similarity analogous to conventional intensity-based image registration. After a ConvNet has been trained with the DLIR framework, it can be used to register pairs of unseen images in one shot. We propose flexible ConvNets designs for affine image registration and for deformable image registration. By stacking multiple of these ConvNets into a larger architecture, we are able to perform coarse-to-fine image registration. We show for registration of cardiac cine MRI and registration of chest CT that performance of the DLIR framework is comparable to conventional image registration while being several orders of magnitude faster.     

GRAM: A framework for geodesic registration on anatomical manifolds

Medical image registration is a challenging problem, especially when there is large anatomical variation in the anatomies. Geodesic registration methods have been proposed to solve the large deformation registration problem. However, analytically defined geodesic paths may not coincide with biologically plausible paths of registration, since the manifold of diffeomorphisms is immensely broader than the manifold spanned by diffeomorphisms between real anatomies. In this paper, we propose a novel framework for large deformation registration using the learned manifold of anatomical variation in the data. In this framework, a large deformation between two images is decomposed into a series of small deformations along the shortest path on an empirical manifold that represents anatomical variation. Using a manifold learning technique, the major variation of the data can be visualized by a low-dimensional embedding, and the optimal group template is chosen as the geodesic mean on the manifold. We demonstrate the advantages of the proposed framework over direct registration with both simulated and real databases of brain images.     

Cross-modal attention for multi-modal image registration

In the past few years, convolutional neural networks (CNNs) have been proven powerful in extracting image features crucial for medical image registration. However, challenging applications and recent advances in computer vision suggest that CNNs are limited in their ability to understand the spatial correspondence between features, which is at the core of image registration. The issue is further exaggerated when it comes to multi-modal image registration, where the appearances of input images can differ significantly. This paper presents a novel cross-modal attention mechanism for correlating features extracted from the multi-modal input images and mapping such correlation to image registration transformation. To efficiently train the developed network, a contrastive learning-based pre-training method is also proposed to aid the network in extracting high-level features across the input modalities for the following cross-modal attention learning. We validated the proposed method on transrectal ultrasound (TRUS) to magnetic resonance (MR) registration, a clinically important procedure that benefits prostate cancer biopsy. Our experimental results demonstrate that for MR-TRUS registration, a deep neural network embedded with the cross-modal attention block outperforms other advanced CNN-based networks with ten times its size. We also incorporated visualization techniques to improve the interpretability of our network, which helps bring insights into the deep learning based image registration methods. The source code of our work is available at https://github.com/DIAL-RPI/Attention-Reg.     

A deep learning framework for hyperspectral image classification using spatial pyramid pooling

In this letter, a new deep learning framework for spectral–spatial classification of hyperspectral images is presented. The proposed framework serves as an engine for merging the spatial and spectral features via suitable deep learning architecture: stacked autoencoders (SAEs) and deep convolutional neural networks (DCNNs) followed by a logistic regression (LR) classifier. In this framework, SAEs is aimed to get useful high-level features for the one-dimensional features which is suitable for the dimension reduction of spectral features, while DCNNs can learn rich features from the training data automatically and has achieved state-of-the-art performance in many image classification databases. Though the DCNNs has shown robustness to distortion, it only extracts features of the same scale, and hence is insufficient to tolerate large-scale variance of object. As a result, spatial pyramid pooling (SPP) is introduced into hyperspectral image classification for the first time by pooling the spatial feature maps of the top convolutional layers into a fixed-length feature. Experimental results with widely used hyperspectral data indicate that classifiers built in this deep learning-based framework provide competitive performance.     

TransMorph: Transformer for unsupervised medical image registration

In the last decade, convolutional neural networks (ConvNets) have been a major focus of research in medical image analysis. However, the performances of ConvNets may be limited by a lack of explicit consideration of the long-range spatial relationships in an image. Recently, Vision Transformer architectures have been proposed to address the shortcomings of ConvNets and have produced state-of-the-art performances in many medical imaging applications. Transformers may be a strong candidate for image registration because their substantially larger receptive field enables a more precise comprehension of the spatial correspondence between moving and fixed images. Here, we present TransMorph, a hybrid Transformer-ConvNet model for volumetric medical image registration. This paper also presents diffeomorphic and Bayesian variants of TransMorph: the diffeomorphic variants ensure the topology-preserving deformations, and the Bayesian variant produces a well-calibrated registration uncertainty estimate. We extensively validated the proposed models using 3D medical images from three applications: inter-patient and atlas-to-patient brain MRI registration and phantom-to-CT registration. The proposed models are evaluated in comparison to a variety of existing registration methods and Transformer architectures. Qualitative and quantitative results demonstrate that the proposed Transformer-based model leads to a substantial performance improvement over the baseline methods, confirming the effectiveness of Transformers for medical image registration.     

DragNet: Learning-based deformable registration for realistic cardiac MR sequence generation from a single frame

Deformable image registration (DIR) can be used to track cardiac motion. Conventional DIR algorithms aim to establish a dense and non-linear correspondence between independent pairs of images. They are, nevertheless, computationally intensive and do not consider temporal dependencies to regulate the estimated motion in a cardiac cycle. In this paper, leveraging deep learning methods, we formulate a novel hierarchical probabilistic model, termed DragNet, for fast and reliable spatio-temporal registration in cine cardiac magnetic resonance (CMR) images and for generating synthetic heart motion sequences. DragNet is a variational inference framework, which takes an image from the sequence in combination with the hidden states of a recurrent neural network (RNN) as inputs to an inference network per time step. As part of this framework, we condition the prior probability of the latent variables on the hidden states of the RNN utilised to capture temporal dependencies. We further condition the posterior of the motion field on a latent variable from hierarchy and features from the moving image. Subsequently, the RNN updates the hidden state variables based on the feature maps of the fixed image and the latent variables. Different from traditional methods, DragNet performs registration on unseen sequences in a forward pass, which significantly expedites the registration process. Besides, DragNet enables generating a large number of realistic synthetic image sequences given only one frame, where the corresponding deformations are also retrieved. The probabilistic framework allows for computing spatio-temporal uncertainties in the estimated motion fields. Our results show that DragNet performance is comparable with state-of-the-art methods in terms of registration accuracy, with the advantage of offering analytical pixel-wise motion uncertainty estimation across a cardiac cycle and being a motion generator. We will make our code publicly available.     

Spectral–spatial classification of hyperspectral images using deep convolutional neural networks

In this letter, a novel deep learning framework for hyperspectral image classification using both spectral and spatial features is presented. The framework is a hybrid of principal component analysis, deep convolutional neural networks (DCNNs) and logistic regression (LR). The DCNNs for hierarchically extract deep features is introduced into hyperspectral image classification for the first time. The proposed technique consists of two steps. First, feature map generation algorithm is presented to generate the spectral and spatial feature maps. Second, the DCNNs-LR classifier is trained to get useful high-level features and to fine-tune the whole model. Comparative experiments conducted over widely used hyperspectral data indicate that DCNNs-LR classifier built in this proposed deep learning framework provides better classification accuracy than previous hyperspectral classification methods.     

ACEnet: Anatomical context-encoding network for neuroanatomy segmentation

Segmentation of brain structures from magnetic resonance (MR) scans plays an important role in the quantification of brain morphology. Since 3D deep learning models suffer from high computational cost, 2D deep learning methods are favored for their computational efficiency. However, existing 2D deep learning methods are not equipped to effectively capture 3D spatial contextual information that is needed to achieve accurate brain structure segmentation. In order to overcome this limitation, we develop an Anatomical Context-Encoding Network (ACEnet) to incorporate 3D spatial and anatomical contexts in 2D convolutional neural networks (CNNs) for efficient and accurate segmentation of brain structures from MR scans, consisting of 1) an anatomical context encoding module to incorporate anatomical information in 2D CNNs and 2) a spatial context encoding module to integrate 3D image information in 2D CNNs. In addition, a skull stripping module is adopted to guide the 2D CNNs to attend to the brain. Extensive experiments on three benchmark datasets have demonstrated that our method achieves promising performance compared with state-of-the-art alternative methods for brain structure segmentation in terms of both computational efficiency and segmentation accuracy.     

A supervised learning framework of statistical shape and probability priors for automatic prostate segmentation in ultrasound images

Prostate segmentation aids in prostate volume estimation, multi-modal image registration, and to create patient specific anatomical models for surgical planning and image guided biopsies. However, manual segmentation is time consuming and suffers from inter-and intra-observer variabilities. Low contrast images of trans rectal ultrasound and presence of imaging artifacts like speckle, micro-calcifications, and shadow regions hinder computer aided automatic or semi-automatic prostate segmentation. In this paper, we propose a prostate segmentation approach based on building multiple mean parametric models derived from principal component analysis of shape and posterior probabilities in a multi-resolution framework. The model parameters are then modified with the prior knowledge of the optimization space to achieve optimal prostate segmentation. In contrast to traditional statistical models of shape and intensity priors, we use posterior probabilities of the prostate region determined from random forest classification to build our appearance model, initialize and propagate our model. Furthermore, multiple mean models derived from spectral clustering of combined shape and appearance parameters are applied in parallel to improve segmentation accuracies. The proposed method achieves mean Dice similarity coefficient value of 0.91 ± 0.09 for 126 images containing 40 images from the apex, 40 images from the base and 46 images from central regions in a leave-one-patient-out validation framework. The mean segmentation time of the procedure is 0.67 ± 0.02 s.     

SatCNN: satellite image dataset classification using agile convolutional neural networks

With the launch of various remote-sensing satellites, more and more high-spatial resolution remote-sensing (HSR-RS) images are becoming available. Scene classification of such a huge volume of HSR-RS images is a big challenge for the efficiency of the feature learning and model training. The deep convolutional neural network (CNN), a typical deep learning model, is an efficient end-to-end deep hierarchical feature learning model that can capture the intrinsic features of input HSR-RS images. However, most published CNN architectures are borrowed from natural scene classification with thousands of training samples, and they are not designed for HSR-RS images. In this paper, we propose an agile CNN architecture, named as SatCNN, for HSR-RS image scene classification. Based on recent improvements to modern CNN architectures, we use more efficient convolutional layers with smaller kernels to build an effective CNN architecture. Experiments on SAT data sets confirmed that SatCNN can quickly and effectively learn robust features to handle the intra-class diversity even with small convolutional kernels, and the deeper convolutional layers allow spontaneous modelling of the relative spatial relationships. With the help of fast graphics processing unit acceleration, SatCNN can be trained within about 40 min, achieving overall accuracies of 99.65% and 99.54%, which is the state-of-the-art for SAT data sets.     

Anatomically constrained and attention-guided deep feature fusion for joint segmentation and deformable medical image registration

The main objective of anatomically plausible results for deformable image registration is to improve model’s registration accuracy by minimizing the difference between a pair of fixed and moving images. Since many anatomical features are closely related to each other, leveraging supervision from auxiliary tasks (such as supervised anatomical segmentation) has the potential to enhance the realism of the warped images after registration. In this work, we employ a Multi-Task Learning framework to formulate registration and segmentation as a joint issue, in which we utilize anatomical constraint from auxiliary supervised segmentation to enhance the realism of the predicted images. First, we propose a Cross-Task Attention Block to fuse the high-level feature from both the registration and segmentation network. With the help of initial anatomical segmentation, the registration network can benefit from learning the task-shared feature correlation and rapidly focusing on the parts that need deformation. On the other hand, the anatomical segmentation discrepancy from ground-truth fixed annotations and predicted segmentation maps of initial warped images are integrated into the loss function to guide the convergence of the registration network. Ideally, a good deformation field should be able to minimize the loss function of registration and segmentation. The voxel-wise anatomical constraint inferred from segmentation helps the registration network to reach a global optimum for both deformable and segmentation learning. Both networks can be employed independently during the testing phase, enabling only the registration output to be predicted when the segmentation labels are unavailable. Qualitative and quantitative results indicate that our proposed methodology significantly outperforms the previous state-of-the-art approaches on inter-patient brain MRI registration and pre- and intra-operative uterus MRI registration tasks within our specific experimental setup, which leads to state-of-the-art registration quality scores of 0.755 and 0.731 (i.e., by 0.8% and 0.5% increases) DSC for both tasks, respectively.     

Multi-scale GCN-assisted two-stage network for joint segmentation of retinal layers and discs in peripapillary OCT images

An accurate and automated tissue segmentation algorithm for retinal optical coherence tomography (OCT) images is crucial for the diagnosis of glaucoma. However, due to the presence of the optic disc, the anatomical structure of the peripapillary region of the retina is complicated and is challenging for segmentation. To address this issue, we develop a novel graph convolutional network (GCN)-assisted two-stage framework to simultaneously label the nine retinal layers and the optic disc. Specifically, a multi-scale global reasoning module is inserted between the encoder and decoder of a U-shape neural network to exploit anatomical prior knowledge and perform spatial reasoning. We conduct experiments on human peripapillary retinal OCT images. We also provide public access to the collected dataset, which might contribute to the research in the field of biomedical image processing. The Dice score of the proposed segmentation network is 0.820 ± 0.001 and the pixel accuracy is 0.830 ± 0.002, both of which outperform those from other state-of-the-art techniques.     

A deep learning framework for pancreas segmentation with multi-atlas registration and 3D level-set

In this paper, we propose and validate a deep learning framework that incorporates both multi-atlas registration and level-set for segmenting pancreas from CT volume images. The proposed segmentation pipeline consists of three stages, namely coarse, fine, and refine stages. Firstly, a coarse segmentation is obtained through multi-atlas based 3D diffeomorphic registration and fusion. After that, to learn the connection feature, a 3D patch-based convolutional neural network (CNN) and three 2D slice-based CNNs are jointly used to predict a fine segmentation based on a bounding box determined from the coarse segmentation. Finally, a 3D level-set method is used, with the fine segmentation being one of its constraints, to integrate information of the original image and the CNN-derived probability map to achieve a refine segmentation. In other words, we jointly utilize global 3D location information (registration), contextual information (patch-based 3D CNN), shape information (slice-based 2.5D CNN) and edge information (3D level-set) in the proposed framework. These components form our cascaded coarse-fine-refine segmentation framework. We test the proposed framework on three different datasets with varying intensity ranges obtained from different resources, respectively containing 36, 82 and 281 CT volume images. In each dataset, we achieve an average Dice score over 82%, being superior or comparable to other existing state-of-the-art pancreas segmentation algorithms.     

Registration, atlas estimation and variability analysis of white matter fiber bundles modeled as currents

This paper proposes a generic framework for the registration, the template estimation and the variability analysis of white matter fiber bundles extracted from diffusion images. This framework is based on the metric on currents for the comparison of fiber bundles. This metric measures anatomical differences between fiber bundles, seen as global homologous structures across subjects. It avoids the need to establish correspondences between points or between individual fibers of different bundles. It can measure differences both in terms of the geometry of the bundles (like its boundaries) and in terms of the density of fibers within the bundle. It is robust to fiber interruptions and reconnections. In addition, a recently introduced sparse approximation algorithm allows us to give an interpretable representation of the fiber bundles and their variations in the framework of currents. First, we used this metric to drive the registration between two sets of homologous fiber bundles of two different subjects. A dense deformation of the underlying white matter is estimated, which is constrained by the bundles seen as global anatomical landmarks. By contrast, the alignment obtained from image registration is driven only by the local gradient of the image. Second, we propose a generative statistical model for the analysis of a collection of homologous bundles. This model consistently estimates prototype fiber bundles (called template), which capture the anatomical invariants in the population, a set of deformations, which align the geometry of the template to that of each subject and a set of residual perturbations. The statistical analysis of both the deformations and the residuals describe the anatomical variability in terms of geometry (stretching, torque, etc.) and "texture" (fiber density, etc.). Third, this statistical modeling allows us to simulate new synthetic bundles according to the estimated variability. This gives a way to interpret the anatomical features that the model detects consistently across the subjects. This may be used to better understand the bias introduced by the fiber extraction methods and eventually to give anatomical characterization of the normal or pathological variability of fiber bundles.     

Tissue self-attention network for the segmentation of optical coherence tomography images on the esophagus

Automatic segmentation of layered tissue is the key to esophageal optical coherence tomography (OCT) image processing. With the advent of deep learning techniques, frameworks based on a fully convolutional network are proved to be effective in classifying pixels on images. However, due to speckle noise and unfavorable imaging conditions, the esophageal tissue relevant to the diagnosis is not always easy to identify. An effective approach to address this problem is extracting more powerful feature maps, which have similar expressions for pixels in the same tissue and show discriminability from those from different tissues. In this study, we proposed a novel framework, called the tissue self-attention network (TSA-Net), which introduces the self-attention mechanism for esophageal OCT image segmentation. The self-attention module in the network is able to capture long-range context dependencies from the image and analyzes the input image in a global view, which helps to cluster pixels in the same tissue and reveal differences of different layers, thus achieving more powerful feature maps for segmentation. Experiments have visually illustrated the effectiveness of the self-attention map, and its advantages over other deep networks were also discussed.