With the increase of the ageing in the world’s population, the ageing and degeneration studies of physiological characteristics in human skin, bones, and muscles become important topics. Research on the ageing of bones, especially the skull, are paid much attention in recent years. In this study, a novel deep learning method representing the ageing-related dynamic attention (ARDA) is proposed. The proposed method can quantitatively display the ageing salience of the bones and their change patterns with age on lateral cephalometric radiographs images (LCR) images containing the craniofacial and cervical spine. An age estimation-based deep learning model based on 14142 LCR images from 4 to 40 years old individuals is trained to extract ageing-related features, and based on these features the ageing salience maps are generated by the Grad-CAM method. All ageing salience maps with the same age are merged as an ARDA map corresponding to that age. Ageing salience maps show that ARDA is mainly concentrated in three regions in LCR images: the teeth, craniofacial, and cervical spine regions. Furthermore, the dynamic distribution of ARDA at different ages and instances in LCR images is quantitatively analyzed. The experimental results on 3014 cases show that ARDA can accurately reflect the development and degeneration patterns in LCR images.

Objective. Ultrasound is the most commonly used examination for the detection and identification of thyroid nodules. Since manual detection is time-consuming and subjective, attempts to introduce machine learning into this process are ongoing. However, the performance of these methods is limited by the low signal-to-noise ratio and tissue contrast of ultrasound images. To address these challenges, we extend thyroid nodule detection from image-based to video-based using the temporal context information in ultrasound videos. Approach. We propose a video-based deep learning model with adjacent frame perception (AFP) for accurate and real-time thyroid nodule detection. Compared to image-based methods, AFP can aggregate semantically similar contextual features in the video. Furthermore, considering the cost of medical image annotation for video-based models, a patch scale self-supervised model (PASS) is proposed. PASS is trained on unlabeled datasets to improve the performance of the AFP model without additional labelling costs. Main results. The PASS model is trained by 92 videos containing 23 773 frames, of which 60 annotated videos containing 16 694 frames were used to train and evaluate the AFP model. The evaluation is performed from the video, frame, nodule, and localization perspectives. In the evaluation of the localization perspective, we used the average precision metric with the intersection-over-union threshold set to 50% (AP@50), which is the area under the smoothed Precision-Recall curve. Our proposed AFP improved AP@50 from 0.256 to 0.390, while the PASS-enhanced AFP further improved the AP@50 to 0.425. AFP and PASS also improve the performance in the valuations of other perspectives based on the localization results. Significance. Our video-based model can mitigate the effects of low signal-to-noise ratio and tissue contrast in ultrasound images and enable the accurate detection of thyroid nodules in real-time. The evaluation from multiple perspectives of the ablation experiments demonstrates the effectiveness of our proposed AFP and PASS models.

Fault intensity diagnosis (FID) plays a pivotal role in monitoring and maintaining mechanical devices within complex industrial systems. As current FID methods are based on chain of thought without considering dependencies among target classes. To capture and explore dependencies, we propose a hierarchical knowledge guided fault intensity diagnosis framework (HKG) inspired by the tree of thought, which is amenable to any representation learning methods. The HKG uses graph convolutional networks to map the hierarchical topological graph of class representations into a set of interdependent global hierarchical classifiers, where each node is denoted by word embeddings of a class. These global hierarchical classifiers are applied to learned deep features extracted by representation learning, allowing the entire model to be end-to-end learnable. In addition, we develop a re-weighted hierarchical knowledge correlation matrix (Re-HKCM) scheme by embedding inter-class hierarchical knowledge into a data-driven statistical correlation matrix (SCM) which effectively guides the information sharing of nodes in graphical convolutional neural networks and avoids over-smoothing issues. The Re-HKCM is derived from the SCM through a series of mathematical transformations. Extensive experiments are performed on four real-world datasets from different industrial domains (three cavitation datasets from SAMSON AG and one existing publicly) for FID, all showing superior results and outperform recent state-of-the-art FID methods.

Since skeletal development is largely completed by adulthood, it is difficult for traditional methods to capture subtle age-related structural changes in bones and surrounding tissues. Recent advances in deep learning have demonstrated remarkable potential in medical image-based age estimation. The cervical vertebrae, as captured in lateral cephalometric radiographs (LCR), have shown particular value in such tasks. To systematically investigate the contribution of different vertebral representations to age estimation, we developed four distinct input modes: (1) Contour (C); (2) Mask (M); (3) Cervical Vertebrae (CV) and (4) Cervical vertebrae region (SR). Using a large-scale LCR dataset of 20,174 subjects aged 4–40 years, grouped into 5-year intervals, we evaluated these modes with deep learning models. The Mean Absolute Error (MAE) was used to evaluate performance. Results indicated that the SR mode achieved the lowest overall MAE, particularly for the C1–C4 combination, followed by CV, while C and M modes showed similar and poorer performance. For subjects younger than 25 years, MAEs for individual vertebrae (C1–2, C3, C4) were less than 5 years across all modes; however, in the 26–40 years group, MAEs for C and M modes exceeded 10 years, whereas CV and SR modes remained below 10 years for most combinations. Combining vertebrae consistently improved accuracy over individual ones, with continuous combinations (e.g., C1–2 + C3) outperforming discontinuous ones (e.g., C1–2 + C4). Visualization of age-related salience revealed that salient regions varied by input mode and expanded with increased information content. These findings underscore the critical importance of incorporating peripheral soft tissue and comprehensive vertebral context for accurate age estimation across a wide age spectrum.

Background: Childhood and adolescence constitute a critical period for craniofacial growth. Understanding its developmental patterns is essential for clinical decision-making in orthodontics and maxillofacial surgery. Traditional cephalometric analysis relies on manual landmarking, which oversimplifies complex morphology and introduces subjectivity. Although deep learning, a key artificial intelligence (AI) technology, has demonstrated remarkable performance in image analysis and classification, most methods still depend on manual annotations during training, perpetuating subjectivity and limiting model generalizability and robustness on large datasets. This hinders the development of objective, comprehensive methods to quantify craniofacial growth that account for its multi-tissue complexity. Methods: To address these limitations, this study developed an end-to-end deep learning framework based on lateral cephalometric radiographs from 41,625 individuals aged 4–18 years. Without relying on manual annotations, the model is designed to autonomously extract dynamic imaging features associated with continuous age intervals in craniofacial development and further discern features related to sexual dimorphism. Gradient-weighted Class Activation Mapping (Grad-CAM) was employed to visualize the learned features, generating population-averaged saliency maps that highlight age-related and sex-related patterns. Furthermore, we introduced two novel quantitative metrics, the Age-related Saliency Index (ASI) and the Sex-related Saliency Index (SSI), to evaluate the significance of developmental and dimorphic characteristics in key craniofacial regions. Results: Age-related saliency maps extended the focus from external contours to internal anatomical details of the bones, intuitively visualizing the relative importance of multiple bone regions during dynamic development, with the ASI providing a quantitative prioritization of these regions. The Sex-related Saliency Index (SSI) quantified the dynamic evolution of sexual dimorphism, demonstrating that early-stage differences were widely distributed across cranial bones and gradually became concentrated in the mandibular region by adulthood. Conclusions: This study established an end-to-end deep learning framework for analyzing large-scale lateral cephalometric radiographs. By generating age- and sex-related average saliency maps and their corresponding quantitative indices, we visualized and quantified the spatiotemporal growth dynamics and sexual dimorphism across distinct craniofacial skeletal regions throughout development. These findings not only validate established developmental theories but also provide novel insights into the coordinated growth patterns of craniofacial bones and sex-specific radiological characteristics, offering clinicians objective quantitative references for assessing developmental stages and guiding the timing of interventions targeting specific craniofacial regions.