arXiv subjects

Junwen Wang

Publications and source records attributed to Junwen Wang.

15 recordsLinked to original sources

arXiv2026

Label tree semantic losses for rich multi-class medical image segmentation

Rich and accurate medical image segmentation is poised to underpin the next generation of AI-defined clinical practice by delineating critical anatomy for pre-operative planning, guiding real-time intra-operative navigation, and supporting precise post-operative assessment. However, commonly used learning methods for medical and surgical imaging segmentation tasks penalise all errors equivalently and thus fail to exploit any inter-class semantics in the label space. This becomes particularly problematic as the cardinality and richness of labels increases to include subtly different classes. In this work, we propose two tree-based semantic loss functions which take advantage of a hierarchical organisation of the labels. We further incorporate our losses in a recently proposed approach for training with sparse, background-free annotations to extend the applicability of our proposed losses. Extensive experiments are reported on two medical and surgical imaging segmentation tasks, namely head MRI for whole brain parcellation with full supervision and neurosurgical hyperspectral imaging for scene understanding with sparse annotations. Results demonstrate consistent improvements over the evaluated task-specific baselines, with the strongest support for the Wasserstein-based compound loss in whole-brain parcellation and for hierarchy-weighted top-level supervision in the sparse HSI setting.

cs.CV↗

arXiv2026

GlobalDentBench: A Multinational Benchmark for Evaluating LLM Clinical Reasoning in Dentistry with Expert Calibration

While large language models (LLMs) hold transformative potential for medicine, their reasoning robustness and safety in real-world clinical scenarios remain critically underexplored, particularly in dentistry. Here we introduce GlobalDentBench, the first multinational dental benchmark, featuring a taxonomy that encompasses 14 dental specialties across 88 countries and regions spanning six continents. The benchmark comprises 8,978 expert-validated questions across three formats (multiple-choice, short-answer, and case-based questions) and assesses three progressive reasoning levels: knowledge recall (L1), routine reasoning (L2), and individualized reasoning (L3). To ensure data quality, the automated construction framework was calibrated by six senior dentists, achieving expert agreement rates of 99.98% for multiple-choice and short-answer questions and 96.78% for the more complex case-based questions. Evaluation of 12 frontier LLMs on GlobalDentBench revealed a sharp, stepwise performance degradation with increasing reasoning complexity. Specifically, accuracy plummeted from 81.34% on multiple-choice to 64.53% on short-answer and 22.34% on case-based questions, while declining markedly from 74.01% at L1 to 55.64% at L2 and 35.71% at L3. More critically, risk analysis of real-world dental cases demonstrated an alarming overall unsafe rate of 31.01% in LLM-generated clinical recommendations, with 4.51% posing risks of irreversible patient harm and risks particularly pronounced in specialties such as orthodontics. These findings expose fundamental limitations in the medical reasoning and safety of current LLMs. Consequently, GlobalDentBench provides a scalable foundation for trustworthy clinical AI evaluation, underscoring the urgent need for rigorous validation before the safe deployment of these models in healthcare.

cs.AI↗

arXiv2026

OOD-SEG: Exploiting out-of-distribution detection techniques for learning image segmentation from sparse multi-class positive-only annotations

Despite significant advancements, segmentation based on deep neural networks in medical and surgical imaging faces several challenges, two of which we aim to address in this work. First, acquiring complete pixel-level segmentation labels for medical images is time-consuming and requires domain expertise. Second, typical segmentation pipelines cannot detect out-of-distribution (OOD) pixels, leaving them prone to spurious outputs during deployment. In this work, we propose a novel segmentation approach which broadly falls within the positive-unlabelled (PU) learning paradigm and exploits tools from OOD detection techniques. Our framework learns only from sparsely annotated pixels from multiple positive-only classes and does not use any annotation for the background class. These multi-class positive annotations naturally fall within the in-distribution (ID) set. Unlabelled pixels may contain positive classes but also negative ones, including what is typically referred to as \emph{background} in standard segmentation formulations. To the best of our knowledge, this work is the first to formulate multi-class segmentation with sparse positive-only annotations as a pixel-wise PU learning problem and to address it using OOD detection techniques. Here, we forgo the need for background annotation and consider these together with any other unseen classes as part of the OOD set. Our framework can integrate, at a pixel-level, any OOD detection approaches designed for classification tasks. To address the lack of existing OOD datasets and established evaluation metric for medical image segmentation, we propose a cross-validation strategy that treats held-out labelled classes as OOD. Extensive experiments on both multi-class hyperspectral and RGB surgical imaging datasets demonstrate the robustness and generalisation capability of our proposed framework.

cs.CV↗

arXiv2025

DentalGPT: Incentivizing Multimodal Complex Reasoning in Dentistry

Reliable interpretation of multimodal data in dentistry is essential for automated oral healthcare, yet current multimodal large language models (MLLMs) struggle to capture fine-grained dental visual details and lack sufficient reasoning ability for precise diagnosis. To address these limitations, we present DentalGPT, a specialized dental MLLM developed through high-quality domain knowledge injection and reinforcement learning. Specifically, the largest annotated multimodal dataset for dentistry to date was constructed by aggregating over 120k dental images paired with detailed descriptions that highlight diagnostically relevant visual features, making it the multimodal dataset with the most extensive collection of dental images to date. Training on this dataset significantly enhances the MLLM's visual understanding of dental conditions, while the subsequent reinforcement learning stage further strengthens its capability for multimodal complex reasoning. Comprehensive evaluations on intraoral and panoramic benchmarks, along with dental subsets of medical VQA benchmarks, show that DentalGPT achieves superior performance in disease classification and dental VQA tasks, outperforming many state-of-the-art MLLMs despite having only 7B parameters. These results demonstrate that high-quality dental data combined with staged adaptation provides an effective pathway for building capable and domain-specialized dental MLLMs.

cs.CV↗

arXiv2025

Analytical Interaction Potentials for Disks in Two Dimensions

Compact analytical forms are derived for the interactions involving thin disks in two dimensions using an integration approach. These include interactions between a disk and a material point, between two disks, and between a disk and a wall. Each object is treated as a continuous medium of materials points interacting by the Lennard-Jones 12-6 potential. By integrating this potential in a pairwise manner, expressions for the potentials and resultant forces between extended objects are obtained. All the results are validated with numerical integrations. The analytical potentials are implemented in LAMMPS and used to simulate two-dimensional suspension of disks with an explicit solvent modeled as a Lennard-Jones liquid. In monodisperse disk suspensions, a disorder-to-order transition of disk packing is observed as the area fraction of disks is increased or as the solvent evaporates. In bidisperse disk suspensions being rapidly dried, stratification is found with the smaller disks enriched at the evaporation front. Such "small-on-top" stratification echoes the similar phenomenon occurring in three-dimensional polydisperse colloidal suspensions that undergo fast drying. These potentials can be applied to a wide range of two-dimensional systems involving disk-like objects.

cond-mat.soft↗

arXiv2025

DeepSeek performs better than other Large Language Models in Dental Cases

Large language models (LLMs) hold transformative potential in healthcare, yet their capacity to interpret longitudinal patient narratives remains inadequately explored. Dentistry, with its rich repository of structured clinical data, presents a unique opportunity to rigorously assess LLMs' reasoning abilities. While several commercial LLMs already exist, DeepSeek, a model that gained significant attention earlier this year, has also joined the competition. This study evaluated four state-of-the-art LLMs (GPT-4o, Gemini 2.0 Flash, Copilot, and DeepSeek V3) on their ability to analyze longitudinal dental case vignettes through open-ended clinical tasks. Using 34 standardized longitudinal periodontal cases (comprising 258 question-answer pairs), we assessed model performance via automated metrics and blinded evaluations by licensed dentists. DeepSeek emerged as the top performer, demonstrating superior faithfulness (median score = 0.528 vs. 0.367-0.457) and higher expert ratings (median = 4.5/5 vs. 4.0/5), without significantly compromising readability. Our study positions DeepSeek as the leading LLM for case analysis, endorses its integration as an adjunct tool in both medical education and research, and highlights its potential as a domain-specific agent.

cs.CL↗

arXiv2025

Tree-based Semantic Losses: Application to Sparsely-supervised Large Multi-class Hyperspectral Segmentation

Hyperspectral imaging (HSI) shows great promise for surgical applications, offering detailed insights into biological tissue differences beyond what the naked eye can perceive. Refined labelling efforts are underway to train vision systems to distinguish large numbers of subtly varying classes. However, commonly used learning methods for biomedical segmentation tasks penalise all errors equivalently and thus fail to exploit any inter-class semantics in the label space. In this work, we introduce two tree-based semantic loss functions which take advantage of a hierarchical organisation of the labels. We further incorporate our losses in a recently proposed approach for training with sparse, background-free annotations. Extensive experiments demonstrate that our proposed method reaches state-of-the-art performance on a sparsely annotated HSI dataset comprising $107$ classes organised in a clinically-defined semantic tree structure. Furthermore, our method enables effective detection of out-of-distribution (OOD) pixels without compromising segmentation performance on in-distribution (ID) pixels.

cs.CV↗

arXiv2024

Extending Structures for Rota-Baxter family Hom-associative Algebras

In this paper, we first define extending datums and unified products of Rota-Baxter family Hom-associative algebras, and theoretically solve the extending structure problem. Moreover, we consider flag datums as an application, and give an example of the extending structure problem. Second, we introduce matched pairs of Rota-Baxter family Hom-associative algebras, and theoretically solve the factorization problem. Finally, we define deformation maps on a Rota-Baxter family Hom extending structure, and theoretically solve the classifying complements problem.

math.RA↗

arXiv2024

Scribble-Based Interactive Segmentation of Medical Hyperspectral Images

Hyperspectral imaging (HSI) is an advanced medical imaging modality that captures optical data across a broad spectral range, providing novel insights into the biochemical composition of tissues. HSI may enable precise differentiation between various tissue types and pathologies, making it particularly valuable for tumour detection, tissue classification, and disease diagnosis. Deep learning-based segmentation methods have shown considerable advancements, offering automated and accurate results. However, these methods face challenges with HSI datasets due to limited annotated data and discrepancies from hardware and acquisition techniques~\cite{clancy2020surgical,studier2023heiporspectral}. Variability in clinical protocols also leads to different definitions of structure boundaries. Interactive segmentation methods, utilizing user knowledge and clinical insights, can overcome these issues and achieve precise segmentation results \cite{zhao2013overview}. This work introduces a scribble-based interactive segmentation framework for medical hyperspectral images. The proposed method utilizes deep learning for feature extraction and a geodesic distance map generated from user-provided scribbles to obtain the segmentation results. The experiment results show that utilising the geodesic distance maps based on deep learning-extracted features achieved better segmentation results than geodesic distance maps directly generated from hyperspectral images, reconstructed RGB images, or Euclidean distance maps.

eess.IV↗

arXiv2024

Extending Structures for Dendriform Algebras

In this paper, we devote to extending structures for dendriform algebras. First, we define extending datums and unified products of dendriform algebras, and theoretically solve the extending structure problem. As an application, we consider flag datums as a special case of extending structures, and give an example of the extending structure problem. Second, we introduce matched pairs and bicrossed products of dendriform algebras and theoretically solve the factorization problem for dendriform algebras. Moreover, we also introduce cocycle semidirect products and nonabelian semidirect products as special cases of unified products. Finally, we define the deformation map on a dendriform extending structure (more general case), not necessary a matched pair, which is more practical in the classifying complements problem.

math.RA↗

arXiv2024

Analytical Interaction Potential for Lennard-Jones Rods

An analytical form has been derived using Ostrogradski's integration method for the interaction between two thin rods of finite lengths in arbitrary relative configurations in a 3-dimensional space, each treated as a line of material points interacting through the Lennard-Jones 12-6 potential. Simplified analytical forms for coplanar, parallel, and collinear rods are also derived. Exact expressions for the force and torque between the rods are obtained. Similar results for a point particle interacting with a thin rod are provided. These interaction potentials can be widely used for analytical descriptions and computational modeling of systems involving rod-like objects such as liquid crystals, colloids, polymers, elongated viruses and bacteria, and filamentous materials including carbon nanotubes, nanowires, biological filaments, and their bundles.

cond-mat.soft↗

arXiv2024

Integrated Lennard-Jones Potential between a Sphere and a Thin Rod

A compact analytical form is derived via an integration approach for the interaction between a sphere and a thin rod of both finite and infinite lengths, with each object treated as a continuous medium of materials points interacting by the Lennard-Jones 12-6 potential. Expressions for the resultant forces and torques are obtained. Various asymptotic limits of the analytical sphere-rod potential are discussed.

cond-mat.soft↗

arXiv2023

Cascade Learning Localises Discriminant Features in Visual Scene Classification

Lack of interpretability of deep convolutional neural networks (DCNN) is a well-known problem particularly in the medical domain as clinicians want trustworthy automated decisions. One way to improve trust is to demonstrate the localisation of feature representations with respect to expert labeled regions of interest. In this work, we investigate the localisation of features learned via two varied learning paradigms and demonstrate the superiority of one learning approach with respect to localisation. Our analysis on medical and natural datasets show that the traditional end-to-end (E2E) learning strategy has a limited ability to localise discriminative features across multiple network layers. We show that a layer-wise learning strategy, namely cascade learning (CL), results in more localised features. Considering localisation accuracy, we not only show that CL outperforms E2E but that it is a promising method of predicting regions. On the YOLO object detection framework, our best result shows that CL outperforms the E2E scheme by $2\%$ in mAP.

cs.CV↗

arXiv2023

A Surface-Based Federated Chow Test Model for Integrating APOE Status, Tau Deposition Measure, and Hippocampal Surface Morphometry

Background: Alzheimer's Disease (AD) is the most common type of age-related dementia, affecting 6.2 million people aged 65 or older according to CDC data. It is commonly agreed that discovering an effective AD diagnosis biomarker could have enormous public health benefits, potentially preventing or delaying up to 40% of dementia cases. Tau neurofibrillary tangles are the primary driver of downstream neurodegeneration and subsequent cognitive impairment in AD, resulting in structural deformations such as hippocampal atrophy that can be observed in magnetic resonance imaging (MRI) scans. Objective: To build a surface-based model to 1) detect differences between APOE subgroups in patterns of tau deposition and hippocampal atrophy, and 2) use the extracted surface-based features to predict cognitive decline. Methods: Using data obtained from different institutions, we develop a surface-based federated Chow test model to study the synergistic effects of APOE, a previously reported significant risk factor of AD, and tau on hippocampal surface morphometry. Results: We illustrate that the APOE-specific morphometry features correlate with AD progression and better predict future AD conversion than other MRI biomarkers. For example, a strong association between atrophy and abnormal tau was identified in hippocampal subregion cornu ammonis 1 (CA1 subfield) and subiculum in e4 homozygote cohort. Conclusion: Our model allows for identifying MRI biomarkers for AD and cognitive decline prediction and may uncover a corner of the neural mechanism of the influence of APOE and tau deposition on hippocampal morphology.

physics.med-ph↗

arXiv2015

Bayesian detection of embryonic gene expression onset in C. elegans

To study how a zygote develops into an embryo with different tissues, large-scale 4D confocal movies of C. elegans embryos have been produced recently by experimental biologists. However, the lack of principled statistical methods for the highly noisy data has hindered the comprehensive analysis of these data sets. We introduced a probabilistic change point model on the cell lineage tree to estimate the embryonic gene expression onset time. A Bayesian approach is used to fit the 4D confocal movies data to the model. Subsequent classification methods are used to decide a model selection threshold and further refine the expression onset time from the branch level to the specific cell time level. Extensive simulations have shown the high accuracy of our method. Its application on real data yields both previously known results and new findings.

stat.AP↗