X-ray imaging machines are widely used in border control checkpoints or public transportation, for luggage scanning and inspection. Recent advances in deep learning enabled automatic object detection of X-ray imaging results to largely reduce labor costs. Compared to tasks on natural images, object detection for X-ray inspection are typically more challenging, due to the varied sizes and aspect ratios of X-ray images, random locations of the small target objects within the redundant background region, etc. In practice, we show that directly applying off-the-shelf deep learning-based detection algorithms for X-ray imagery can be highly time-consuming and ineffective. To this end, we propose a Task-Driven Cropping scheme, dubbed TDC, for improving the deep image detection algorithms towards efficient and effective luggage inspection via X-ray images. Instead of processing the whole X-ray images for object detection, we propose a two-stage strategy, which first adaptively crops X-ray images and only preserves the task-related regions, i.e., the luggage regions for security inspection. A task-specific deep feature extractor is used to rapidly identify the importance of each X-ray image pixel. Only the regions that are useful and related to the detection tasks are kept and passed to the follow-up deep detector. The varied-scale X-ray images are thus reduced to the same size and aspect ratio, which enables a more efficient deep detection pipeline. Besides, to benchmark the effectiveness of X-ray image detection algorithms, we propose a novel dataset for X-ray image detection, dubbed SIXray-D, based on the popular SIXray dataset. In SIXray-D, we provide the complete and more accurate annotations of both object classes and bounding boxes, which enables model training for supervised X-ray detection methods. Our results show that our proposed TDC algorithm can effectively boost popular detection algorithms, by achieving better detection mAPs or reducing the run time.

Automated Optical Inspection (AOI) is widely used by manufacturers for the detection of defects in Printed Circuit boards (PCBs). Recent works have proposed to apply deep learning for defect detection, which is much faster and cheaper than manual inspection. However, AOI can only capture defects on the outmost layers of PCBs using cameras, while modern high-speed circuit PCBs usually have multiple internal layers that need to be inspected. Compared to optical sensors, X-ray tomography provides non-invasive imaging results of all PCB layers. Though one can directly apply an off-the-shelf deep detection model trained on optical domains for X-ray imagery, we show that it usually leads to much lower accuracies in practice. The degraded performance is mainly due to the relatively low quality of X-ray imaging results and the gaps between optical and X-ray modalities. Furthermore, no X-ray PCB image dataset is publicly available for training deep defect detectors. To this end, we propose a novel dataset for X-ray PCB defect detection, dubbed XD-PCB. In XD-PCB, we provide a benchmark for training X-ray automated defect detection models containing synthesized X-ray images and real X-ray images with real defects. However, in a practical environment, retraining the deep model for every unseen X-ray domain is inefficient due to the domain gaps created by different X-ray machine settings and the scarcity of defects. Thus, we propose a domain adaptation framework, dubbed FDX, to improve the efficiency of X-ray PCB defect detection methods. By minimizing the differences between the deep features extracted from abundant training images and the scarce unseen images, we improve the model’s performance in a practical situation, thus enhancing the generalization ability and efficiency of deep detection algorithms when exposed to unseen domains. Our results demonstrate that XD-PCB provides a valuable training baseline for X-ray PCB defect detection, and our proposed FDX framework can effectively increase the popular deep learning model by achieving an increment of 10% in terms of APs compared to other adaptation methods.

Objective. Electrocardiogram (ECG) analysis is vital for the diagnosis of cardiac conditions and monitoring human physiological states. However, challenges such as signal perturbations, inconsistent quality, and signal inference undermine the reliability of ECG analysis. Despite advances in large language models (LLMs), their application in enhancing ECG-based physiological measurements remains underexplored. To address these challenges, the objective is to develop a novel multimodal framework that integrates ECG signals with textual instructions for robust denoising and signal quality assessment, enabling effective physiological analysis across diverse tasks. Approach. We propose cross-modal attention (CMA)-ECG, a multimodal framework that employs a hybrid cross-attention mechanism to align ECG and text features with task-specific heads, combined with a pre-trained LLM for contextual enhancement. The framework leverages pretrained LLMs with 7B parameters, balancing accuracy and computational requirements for practical deployment. Main results. Extensive experiments on multiple datasets demonstrate that CMA-ECG achieves state-of-the-art (SOTA) performance in robustness to perturbations, quality assessment, and denoising. CMA-ECG achieves up to 8.8% higher area under the ROC curve in quality assessment and 20% lower mean squared error in denoising compared to SOTA baselines, ensuring reliable ECG processing. Significance. This approach advances ECG analysis by integrating signal and contextual data, offering a robust solution for physiological monitoring and analysis, ensuring reliable ECG processing.

Objective: Accurate physiological assessment of cardiac function from heart sounds remains challenging due to background noise, variable heart rates, and the need for reliable cardiac-cycle segmentation. This study aimed to develop a fully E2E deep learning framework that extracts diagnostic information directly from raw heart sound recordings for cardiac abnormality detection and classification. Approach: We propose HS-MMNet, an E2E multi-modal deep learning framework designed for physiological heart sound analysis. Recordings are preprocessed (normalization and 25–400 Hz bandpass filtering) and divided into fixed-length 2.5-s segments. A Convolution Head with multi-atrous spatial pyramid and channel-spatial attention extracts fine-grained local temporal patterns from the filtered 1-D waveform. A Transformer Head captures long-range spectro-temporal dependencies from Log-Mel spectrograms. These hypotheses are iteratively fused by a novel Multi-Hypothesis Cross-Attention (MH-CA) module with cyclic query-key-value assignment and a hypothesis-mixing MLP, enabling rich cross-site interaction and effective suppression of noise and non-informative regions. Recording-level classification is obtained via a fully connected layer. Main results: On the PhysioNet/CinC Challenge 2016 dataset, HS-MMNet achieved 94.80% accuracy, 92.10% sensitivity, 96.85% specificity, 87.50% precision, and 89.74% F1-score, outperforming all previously reported methods. On the balanced five-class Yaseen dataset (normal, aortic stenosis, mitral regurgitation, mitral stenosis, mitral valve prolapse), it attained 99.60% macro-averaged precision, recall, and F1-score with only four misclassifications in 1000 recordings, establishing new state-of-the-art (SOTA) benchmarks. Significance: HS-MMNet represents an advance in automated physiological measurement from heart sounds. By eliminating cardiac cycle detection and multi-channel requirements while achieving SOTA diagnostic performance, it provides a practical, scalable solution for accurate cardiovascular screening with primary-care and low-resource settings.

Video anomaly detection (VAD) has evolved from handcrafted feature-based methods to deep learning and foundation model-based approaches, enabling improved detection of complex and previously unseen events. This review examines this evolution through four paradigms: closed-set, open-set, open-vocabulary, and open-world VAD. A unified taxonomy is developed to organize representative methods according to learning paradigms, model architectures, supervision strategies, and semantic capabilities. The review further compares benchmark datasets, evaluation protocols, and performance metrics, with emphasis on the challenges of detecting and recognizing unseen anomalies in realistic environments. Recent advances in vision–language models (VLMs) are discussed in terms of their ability to incorporate language knowledge and open-ended semantic reasoning into VAD. Key challenges related to generalization, computational efficiency, multimodal understanding, and real-world deployment are identified. Finally, emerging directions involving large language models (LLMs), multimodal large language models (MLLMs), and video large language models (Video-LLMs), together with agentic AI and autonomous reasoning, are discussed toward more adaptive and interpretable VAD systems.