Deformable medical image registration is essential in medical image analyses. Recent transformer-based registration methods have achieved high registration accuracy. However, these methods often rely on patch embedding at the beginning of encoding, resulting in limited ability to capture detailed anatomical structural information in the images and explore local semantic relationships within individual patches. Here, we proposed a novel Dual-feet Encoder (DFEnc) to asynchronously model semantic information from moving and fixed images at various scales through two separate branches in three steps. For each step, features from adjacent resolution levels were processed by a Single Step Hybrid Extractor (SSHExt), which performed patch convolution to preserve local information, followed by several transformer blocks to capture global context. Dense connections were employed to enhance semantic awareness across adjacent feature resolution levels. Additionally, we introduced a Feature Fusion-based Decoder (FFDec) to progressively fuse features related to the fixed and moving images and to generate intermediate deformation fields at each stage, enabling accurate image alignment through stepwise warping and alignment refinement. Extensive ablation studies demonstrated the effectiveness of the proposed DFEnc, SSHExt, and FFDec. Compared to a state-of-the-art AutoFuse-Trans method, our approach yielded improvements in Dice of 1.14%, 1.77%, and 4.47% on the ACDC, OASIS, and Abdomen CT datasets, respectively, while maintaining relatively low computational cost. These results suggest the utility of the proposed approach for broad research and clinical applications.

Cardiovascular health is vital to human well-being, and cardiac magnetic resonance (CMR) imaging is considered the clinical reference standard for diagnosing cardiovascular disease. However, its adoption is hindered by long scan times, complex contrasts, and inconsistent quality. While deep learning methods perform well on specific CMR imaging sequences, they often fail to generalize across modalities and sampling schemes. The lack of benchmarks for high-quality, fast CMR image reconstruction further limits technology comparison and adoption. The CMRxRecon2024 challenge, attracting over 200 teams from 18 countries, addressed these issues with two tasks: generalization to unseen modalities and robustness to diverse undersampling patterns. We introduced the largest public multi-modality CMR raw dataset, an open benchmarking platform, and shared code. Analysis of the best-performing solutions revealed that prompt-based adaptation and enhanced physics-driven consistency enabled strong cross-scenario performance. These findings establish principles for generalizable reconstruction models and advance clinically translatable AI in cardiovascular imaging.

Image reconstruction from undersampled k-space data provides a way for accelerated cardiac magnetic resonance imaging (MRI). However, difficulties in acquiring fully sampled data and rapid respiratory and cardiac motion indicate a dire need for unsupervised reconstruction from highly undersampled acquisitions. In this work, we propose a dual-domain parallel unrolled reconstruction network (DDPUR-Net) that encompasses an image-domain unrolled reconstruction network and a parallel k-space reconstruction branch for highly accelerated cardiac MRI reconstruction. The proposed framework is designed to integrate the local anatomical details provided by the imagedomain and the global information provided by the k -space data. In addition, we introduce a novel information interaction mechanism that fuses the features from both domains and passes the fused features to the two reconstruction branches, enabling utilization of the complementary information from both domains for enhanced reconstruction. For two private and public 3D and 2D cardiac MRI datasets, the proposed approach achieved promising reconstruction accuracy and outperformed several widely used comparative methods.

Cardiac motion estimation plays a crucial role in the diagnosis of cardiovascular diseases, which pose a serious threat to human health. Deformable registration-based methods have shown great potential for directly estimating deformations without relying on explicit feature extraction. However, general registration methods leverage a single pixel value to represent the displacement at each voxel, which limits their effectiveness for cardiac motion estimation. Since cardiac motion includes diverse sources and varying magnitudes, this single representation may lead to misalignment. To mitigate this issue, we propose a novel approach that explicitly decompose deformations into multiple components and assign different weights to each component, thereby achieving precise fitting of complex cardiac motion. The proposed method incorporates a five-stage Feature Mixerbased Decoder (FMDec) to perform progressive registration. At each stage of FMDec, a Feature Mixer Transformer (FMTrans) establishes an explicit mapping between fixed/moving features and the estimated deformation field. Within FMTrans, motion decomposition and component-wise adaptive learning are performed through a multi-head cross-attention mechanism and a Component-wise Weighting Summation (CWS) module. Experiments on the cardiac MRI datasets, ACDC and CMRxMotion, demonstrated that our method improved Dice scores by 1.1 % and 1.9 % compared to AutoFuse-Trans, respectively.

Boundary value analysis is a typical conventional testing technique. However, manually identifying input regions and writing test cases are labor-intensive and time-consuming. In this paper, we propose a search-based random testing approach, which automatically generates test data along the boundaries of semantic regions of the input domain. The experiments on mutated programs confirm the effectiveness and efficiency of the proposed approach. Furthermore, our approach significantly outperforms the conventional ART (Adaptive Random Testing) methods, which sample test cases evenly across the input regions. Our approach also outperforms EvoSuite, a state-of-the-art tool that generates test cases satisfying certain coverage criterion.

Debugging has long been recognized as one of the most labour- and time- consuming activities in software development. Recent research on automated debugging tries to facilitate this process by automatically generating patches for buggy programs so that they pass a predefined test suite. Despite the promising experimental results, several major obstacles emerge when we apply these techniques in active coding process. Inadequate test cases, multiple errors in one program and possible bug categories overlooked by existing fix generation search spaces impede these techniques to perform at their best.

Effective verification methods are necessary for finding bugs in complex system design. Given domain specific knowledge, design verification engineers typically specify cross coverage for certain risky areas where bugs tend to appear. This paper proposes a pragmatic coverage model based on cross coverage. We address the verification on user specified cross coverage regions. The proposed analysis models the probability of exposing the bug within a given number of samplings, and derives the expected number of samples until bug detection. The approach is applicable to random, round-robin, and hybrid sampling strategies in recurring and nonrecurring cases based on our cross coverage model. We have written Matlab and C programs that use our formulas to calculate the probabilities. Experimental results show that our analysis is consistent with Monte Carlo simulation.

Opposing to the oracle assumption, a trustworthy test oracle is not always available in real practice. Since manually written oracles and human judgements are still widely used, testers and programmers are in fact facing a high risk of erroneous test oracles. However, test oracle errors can bring much confusion thus causing extra time consumption in the debugging process. As substantiated by our experiment on the Siemens Test Suite, automatic fault localization algorithms suffer severely from erroneous test oracles, which impede them from reducing debugging time to the full extent. This paper proposes a simple but effective approach to debug the test oracle. Based on the observation that test cases covering similar lines of code usually generate similar results, we are able to identify suspicious test cases that are differently judged by the test oracle from their neighbors. To validate the effectiveness of our approach, experiments are conducted on both the Siemens Test Suite and grep. The results show that averagely over 75 percent of the highlighted test cases are actually test oracle errors. Moreover, performance of fault localization algorithms recovered remarkably with the debugged oracles.