Large deformations of organs, caused by diverse shapes and nonlinear shape changes, pose a significant challenge for medical image registration. Traditional registration methods need to iteratively optimize an objective function via a specific deformation model along with meticulous parameter tuning, but which have limited capabilities in registering images with large deformations. While deep learning-based methods can learn the complex mapping from input images to their respective deformation field, it is regression-based and is prone to be stuck at local minima, particularly when large deformations are involved. To this end, we present Stochastic Planner-Actor-Critic (spac), a novel reinforcement learning-based framework that performs step-wise registration. The key notion is warping a moving image successively by each time step to finally align to a fixed image. Considering that it is challenging to handle high dimensional continuous action and state spaces in the conventional reinforcement learning (RL) framework, we introduce a new concept `Plan' to the standard Actor-Critic model, which is of low dimension and can facilitate the actor to generate a tractable high dimensional action. The entire framework is based on unsupervised training and operates in an end-to-end manner. We evaluate our method on several 2D and 3D medical image datasets, some of which contain large deformations. Our empirical results highlight that our work achieves consistent, significant gains and outperforms state-of-the-art methods.

Large training datasets almost always contain examples with inaccurate or incorrect labels. Deep Neural Networks (DNNs) tend to overfit training label noise, resulting in poorer model performance in practice. To address this problem, we propose a robust label correction framework combining the ideas of small loss selection and noise correction, which learns network parameters and reassigns ground truth labels iteratively. Taking the expertise of DNNs to learn meaningful patterns before fitting noise, our framework first trains two networks over the current dataset with small loss selection. Based on the classification loss and agreement loss of two networks, we can measure the confidence of training data. More and more confident samples are selected for label correction during the learning process. We demonstrate our method on both synthetic and real-world datasets with different noise types and rates, including CIFAR-10, CIFAR-100 and Clothing1M, where our method outperforms the baseline approaches.

In this paper, we propose a new few-shot object detection (FSOD) framework that introduces a new contrastive branch to extract the class representation of images, which improves the generalization performance of the detection model for novel classes. Additionally, we investigate the effectiveness of both self-supervised and supervised contrastive losses for class-specific encoding in our framework. Experimental results on the benchmark datasets indicate that our proposed method archives the state-of-the-art performance compared with existing FSOD methods.

Abstract Radiation therapy is the standard treatment for early stage Nasopharyngeal cancer (NPC). Thus, accurate delineation of target volumes at risk in NPC is important. While manual delineation is time‐consuming and labour‐intensive process and also leads to significant inter‐ and intra‐practitioner variability. Thus, computer‐aided segmentation algorithm is required. However, segmentation task is not trivial due to large variations (e.g., shape and size) of nasopharynx structure across subjects. Moreover, extreme foreground and background class imbalance in NPC segmentation remains challenge. In this paper, we propose a threedimensional densely connected convolutional neural network with multi‐scale feature pyramids for NPC segmentation. We adapt the densely connected convolutional block into a new structure via adding feature pyramids. The concatenated pyramid feature carries multi‐scale and hierarchical semantic information which is effective for segmenting different size of tumors and perceiving hierarchical context information. To address the foreground and background imbalance problem, we propose an enhanced version of focal loss. It prevents the large number of negative voxels far from boundaries from overwhelming the segmentation algorithm. We validated the proposed method on 120 clinical subjects. Experimental results demonstrate that our approach out‐performed state‐of‐the‐art methods and human experts.

For iris recognition in non-cooperative environments, iris segmentation has been regarded as the first most important challenge still open to the biometric community, affecting all downstream tasks from normalization to recognition. In recent years, deep learning technologies have gained significant popularity among various computer vision tasks and also been introduced in iris biometrics, especially iris segmentation. To investigate recent developments and attract more interest of researchers in the iris segmentation method, we organized the 2021 NIR Iris Challenge Evaluation in Non-cooperative Environments: Segmentation and Localization (NIR-ISL 2021) at the 2021 International Joint Conference on Biometrics (IJCB 2021). The challenge was used as a public platform to assess the performance of iris segmentation and localization methods on Asian and African NIR iris images captured in non-cooperative environments. The three best-performing entries achieved solid and satisfactory iris segmentation and localization results in most cases, and their code and models have been made publicly available for reproducibility research.

Fooling people with highly realistic fake images generated with Deepfake or GANs brings a great social disturbance to our society. Many methods have been proposed to detect fake images, but they are vulnerable to adversarial perturbations – intentionally designed noises that can lead to the wrong prediction. Existing methods of attacking fake image detectors usually generate adversarial perturbations to perturb almost the entire image. This is redundant and increases the perceptibility of perturbations. In this paper, we propose a novel method to disrupt the fake image detection by determining key pixels to a fake image detector and attacking only the key pixels, which results in the L0and the L2norms of adversarial perturbations much less than those of existing works. Experiments on two public datasets with three fake image detectors indicate that our proposed method achieves state-of the-art performance in both white-box and black-box attacks.

Deep neural networks have been demonstrated to be vulnerable to adversarial attacks: subtle perturbation can completely change prediction result. The vulnerability has led to a surge of research in this direction, including adversarial attacks on object detection networks. However, previous studies are dedicated to attacking anchor-based object detectors. In this paper, we present the first adversarial attack on anchor-free object detectors. It conducts category-wise, instead of previously instance-wise, attacks on object detectors, and leverages high-level semantic information to efficiently generate transferable adversarial examples, which can also be transferred to attack other object detectors, even anchor-based detectors such as Faster R-CNN. Experimental results on two benchmark datasets demonstrate that our proposed method achieves state-of-the-art performance and transferability.

Training a model-free deep reinforcement learning model to solve image-to-image translation is difficult since it involves high-dimensional continuous state and action spaces. In this paper, we draw inspiration from the recent success of the maximum entropy reinforcement learning framework designed for challenging continuous control problems to develop stochastic policies over high dimensional continuous spaces including image representation, generation, and control simultaneously. Central to this method is the Stochastic Actor-Executor-Critic (SAEC) which is an off-policy actor-critic model with an additional executor to generate realistic images. Specifically, the actor focuses on the high-level representation and control policy by a stochastic latent action, as well as explicitly directs the executor to generate low-level actions to manipulate the state. Experiments on several image-to-image translation tasks have demonstrated the effectiveness and robustness of the proposed SAEC when facing high-dimensional continuous space problems.

The deep neural network is vulnerable to adversarial examples. Adding imperceptible adversarial perturbations to images is enough to make them fail. Most existing research focuses on attacking image classifiers or anchor-based object detectors, but they generate globally perturbation on the whole image, which is unnecessary. In our work, we leverage higher-level semantic information to generate high aggressive local perturbations for anchor-free object detectors. As a result, it is less computationally intensive and achieves a higher black-box attack as well as transferring attack performance. The adversarial examples generated by our method are not only capable of attacking anchor-free object detectors, but also able to be transferred to attack anchor-based object detector.

For dilated cardiomyopathy (DCM) patients, immediate emergency diagnosis and treatment are critical for life saving and later recovery.T1 mapping is a non-invasive and effective diagnostic imaging approach to detect DCM.However, it is a demanding and time-consuming approach.In this paper, we propose an attention-based network structure, which can automatically identify DCM patients in a speedy manner to prioritize their treatment.In the proposed method, we adopt attention modules to generate attention-aware features.Inside each attention module, a bottom-up top-down feed-forward structure is used to unfold the feed-forward and feed-back attention processes into a single feed-forward process.It allows the network to focus more on determining useful information about the current output that is significant in the input data.Moreover, inspired by the residual network idea, we make full use of the characteristics of the original data.Combined residual block, we design down-residual modules for classification tasks.It consists of seven convolution layers and three layers of residual blocks.Our network achieves the most advanced recognition performance on cardiac datasets.We evaluated our approach on CMR(cardiac magnetic resonance) T1 mapping images with lower PSNR(peak signal to noise ratio), and the results demonstrate that our architecture outperforms previous approaches.

Segmentation of multiple surfaces in medical images is a challenging problem, further complicated by the frequent presence of weak boundary evidence, large object deformations, and mutual influence between adjacent objects. This paper reports a novel approach to multi-object segmentation that incorporates both shape and context prior knowledge in a 3-D graph-theoretic framework to help overcome the stated challenges. We employ an arc-based graph representation to incorporate a wide spectrum of prior information through pair-wise energy terms. In particular, a shape-prior term is used to penalize local shape changes and a context-prior term is used to penalize local surface-distance changes from a model of the expected shape and surface distances, respectively. The globally optimal solution for multiple surfaces is obtained by computing a maximum flow in a low-order polynomial time. The proposed method was validated on intraretinal layer segmentation of optical coherence tomography images and demonstrated statistically significant improvement of segmentation accuracy compared to our earlier graph-search method that was not utilizing shape and context priors. The mean unsigned surface positioning errors obtained by the conventional graph-search approach (6.30 ±1.58 μ m) was improved to 5.14±0.99 μ m when employing our new method with shape and context priors.

Positron emission tomography (PET)-computed tomography (CT) images have been widely used in clinical practice for radiotherapy treatment planning of the radiotherapy. Many existing segmentation approaches only work for a single imaging modality, which suffer from the low spatial resolution in PET or low contrast in CT. In this work, we propose a novel method for the co-segmentation of the tumor in both PET and CT images, which makes use of advantages from each modality: the functionality information from PET and the anatomical structure information from CT. The approach formulates the segmentation problem as a minimization problem of a Markov random field model, which encodes the information from both modalities. The optimization is solved using a graph-cut based method. Two sub-graphs are constructed for the segmentation of the PET and the CT images, respectively. To achieve consistent results in two modalities, an adaptive context cost is enforced by adding context arcs between the two sub-graphs. An optimal solution can be obtained by solving a single maximum flow problem, which leads to simultaneous segmentation of the tumor volumes in both modalities. The proposed algorithm was validated in robust delineation of lung tumors on 23 PET-CT datasets and two head-and-neck cancer subjects. Both qualitative and quantitative results show significant improvement compared to the graph cut methods solely using PET or CT.

Many computer vision applications can be formulated as labeling problems. However, multilabeling problems are usually very challenging to solve, especially when some ordering constraints are enforced. We solve in this paper a five-parts labeling problem proposed in [6, 7]. In this model, one wants to find an optimal labeling for an image with five possible parts: “left”, “right”, “top”, “bottom” and “center”. The geometric ordering constraints can be read naturally from the names. No previous method can solve the problem with globally optimal solutions in a linear space complexity. We propose an efficient dynamic programming based algorithm which guarantees the global optimal labeling for the five-parts model. The time complexity is O(N1.5) and the space complexity is O(N), with N being the number of pixels in the image. In practice, it runs faster than previous methods. Moreover, it works for both 4-neighborhood and 8-neighborhood settings, and can be easily parallelized for GPU.

In this paper, we propose a novel method to reduce the magnitude of 4D CT artifacts by stitching two images with a data-driven regularization constrain, which helps preserve the local anatomy structures. Our method first computes an interface seam for the stitching in the overlapping region of the first image, which passes through the "smoothest" region, to reduce the structure complexity along the stitching interface. Then, we compute the displacements of the seam by matching the corresponding interface seam in the second image. We use sparse 3D features as the structure cues to guide the seam matching, in which a regularization term is incorporated to keep the structure consistency. The energy function is minimized by solving a multiple-label problem in Markov Random Fields with an anatomical structure preserving regularization term. The displacements are propagated to the rest of second image and the two image are stitched along the interface seams based on the computed displacement field. The method was tested on both simulated data and clinical 4D CT images. The experiments on simulated data demonstrated that the proposed method was able to reduce the landmark distance error on average from 2.9 mm to 1.3 mm, outperforming the registration-based method by about 55%. For clinical 4D CT image data, the image quality was evaluated by three medical experts, and all identified much fewer artifacts from the resulting images by our method than from those by the compared method.

This paper proposes an algorithm to measure the width of retinal vessels in fundus photographs using graph-based algorithm to segment both vessel edges simultaneously. First, the simultaneous two-boundary segmentation problem is modeled as a two-slice, 3-D surface segmentation problem, which is further converted into the problem of computing a minimum closed set in a node-weighted graph. An initial segmentation is generated from a vessel probability image. We use the REVIEW database to evaluate diameter measurement performance. The algorithm is robust and estimates the vessel width with subpixel accuracy. The method is used to explore the relationship between the average vessel width and the distance from the optic disc in 600 subjects.

Multiple surface searching with only image intensity information is a difficult job in the presence of high noise and weak edges. We present in this paper a novel method for globally optimal multi-surface searching with a shape prior represented by convex pairwise energies. A 3-D graph-theoretic framework is employed. An arc-weighted graph is constructed based on a shape model built from training datasets. A wide spectrum of constraints is then incorporated. The shape prior term penalizes the local topological change from the original shape model. The globally optimal solution for multiple surfaces can be obtained by computing a maximum flow in low-order polynomial time. Compared with other graph-based methods, our approach provides more local and flexible control of the shape. We also prove that our algorithm can handle the detection of multiple crossing surfaces with no shared voxels. Our method was applied to several application problems, including medical image segmentation, scenic image segmentation, and image resizing. Compared with results without using shape prior information, our improvement was quite impressive, demonstrating the promise of our method.