Rongzhao Zhang

Chao Ding, Mouxiao Bian, Tianbin Li et al.

Large language models(LLMs) increasingly match expert performance on licensing examinations, yet routine clinical use remains limited because governance requires auditable reasoning, safety and ethics alignment, and resilience to adversarial misuse. Here we present SafeMed-R1, trained with a traceable Clinical Trust Signals(CTS) pipeline that links each reasoning instance to clinician rubric scores and edit histories, and aligned through safety and ethics supervision and red team stress testing. SafeMed-R1 attains a macro-averaged accuracy of 79.6% across clinical benchmarks. Under adversarial safety testing, it shows the lowest aggregated risk and reduces unsafe outputs by about 3 to 5% relative to its baseline. In a paired expert study of 30 medication safety vignettes, SafeMed-R1 matches PGY1 and PGY2 residents on medical correctness and scores higher for medication safety, guideline consistency, and clinical usefulness. Collectively, these results suggest that clinician-audited supervision provenance, together with domain-tailored safety and ethics alignment, can strengthen governance-relevant evidence without relying on inference-time retrieval or citation grounding.

Hang Ding, Peidong Liu, Junqiao Wang et al.

The development of autonomous web agents, powered by Large Language Models (LLMs) and reinforcement learning (RL), represents a significant step towards general-purpose AI assistants. However, training these agents is severely hampered by the challenges of interacting with the live internet, which is inefficient, costly, and fraught with risks. Model-based reinforcement learning (MBRL) offers a promising solution by learning a world model of the environment to enable simulated interaction. This paper introduces DynaWeb, a novel MBRL framework that trains web agents through interacting with a web world model trained to predict naturalistic web page representations given agent actions. This model serves as a synthetic web environment where an agent policy can dream by generating vast quantities of rollout action trajectories for efficient online reinforcement learning. Beyond free policy rollouts, DynaWeb incorporates real expert trajectories from training data, which are randomly interleaved with on-policy rollouts during training to improve stability and sample efficiency. Experiments conducted on the challenging WebArena and WebVoyager benchmarks demonstrate that DynaWeb consistently and significantly improves the performance of state-of-the-art open-source web agent models. Our findings establish the viability of training web agents through imagination, offering a scalable and efficient way to scale up online agentic RL.

Rongzhao Zhang, Junqiao Wang, Shuyun Yang et al.

Multimodal clinical reasoning in the field of gastrointestinal (GI) oncology necessitates the integrated interpretation of endoscopic imagery, radiological data, and biochemical markers. Despite the evident potential exhibited by Multimodal Large Language Models (MLLMs), they frequently encounter challenges such as context dilution and hallucination when confronted with intricate, heterogeneous medical histories. In order to address these limitations, a hierarchical Multi-Agent Framework is proposed, which emulates the collaborative workflow of a human Multidisciplinary Team (MDT). The system attained a composite expert evaluation score of 4.60/5.00, thereby demonstrating a substantial improvement over the monolithic baseline. It is noteworthy that the agent-based architecture yielded the most substantial enhancements in reasoning logic and medical accuracy. The findings indicate that mimetic, agent-based collaboration provides a scalable, interpretable, and clinically robust paradigm for automated decision support in oncology.

Yuan Tian, Kaiyuan Ji, Rongzhao Zhang et al.

Medical image re-identification (MedReID) is under-explored so far, despite its critical applications in personalized healthcare and privacy protection. In this paper, we introduce a thorough benchmark and a unified model for this problem. First, to handle various medical modalities, we propose a novel Continuous Modality-based Parameter Adapter (ComPA). ComPA condenses medical content into a continuous modality representation and dynamically adjusts the modality-agnostic model with modality-specific parameters at runtime. This allows a single model to adaptively learn and process diverse modality data. Furthermore, we integrate medical priors into our model by aligning it with a bag of pre-trained medical foundation models, in terms of the differential features. Compared to single-image feature, modeling the inter-image difference better fits the re-identification problem, which involves discriminating multiple images. We evaluate the proposed model against 25 foundation models and 8 large multi-modal language models across 11 image datasets, demonstrating consistently superior performance. Additionally, we deploy the proposed MedReID technique to two real-world applications, i.e., history-augmented personalized diagnosis and medical privacy protection. Codes and model is available at \href{https://github.com/tianyuan168326/All-in-One-MedReID-Pytorch}{https://github.com/tianyuan168326/All-in-One-MedReID-Pytorch}.

Run Xu, Lu Li, Rongzhao Zhang et al.

Recent multimodal large language models have shown promising ability in generating humorous captions for images, yet they still lack stable control over explicit cultural context, making it difficult to jointly maintain image relevance, contextual appropriateness, and humor quality under a specified cultural background. To address this limitation, we introduce a new multimodal generation task, culture-aware humorous captioning, which requires a model to generate a humorous caption conditioned on both an input image and a target cultural context. Captions generated under different cultural contexts are not expected to share the same surface form, but should remain grounded in similar visual situations or humorous rationales.To support this task, we establish a six-dimensional evaluation framework covering image relevance, contextual fit, semantic richness, reasonableness, humor, and creativity. We further propose a staged alignment framework that first initializes the model with high-resource supervision under the Western cultural context, then performs multi-dimensional preference alignment via judge-based GRPO with a Degradation-aware Prototype Repulsion Constraint to mitigate reward hacking in open-ended generation, and finally adapts the model to the Eastern cultural context with a small amount of supervision. Experimental results show that our method achieves stronger overall performance under the proposed evaluation framework, with particularly large gains in contextual fit and a better balance between image relevance and humor under cultural constraints.

Rongzhao Zhang, Han Zhang, Albert C. S. Chung

Automatic mammogram classification and mass segmentation play a critical role in a computer-aided mammogram screening system. In this work, we present a unified mammogram analysis framework for both whole-mammogram classification and segmentation. Our model is designed based on a deep U-Net with residual connections, and equipped with the novel hybrid deep supervision (HDS) scheme for end-to-end multi-task learning. As an extension of deep supervision (DS), HDS not only can force the model to learn more discriminative features like DS, but also seamlessly integrates segmentation and classification tasks into one model, thus the model can benefit from both pixel-wise and image-wise supervisions. We extensively validate the proposed method on the widely-used INbreast dataset. Ablation study corroborates that pixel-wise and image-wise supervisions are mutually beneficial, evidencing the efficacy of HDS. The results of 5-fold cross validation indicate that our unified model matches state-of-the-art performance on both mammogram segmentation and classification tasks, which achieves an average segmentation Dice similarity coefficient (DSC) of 0.85 and a classification accuracy of 0.89. The code is available at https://github.com/angrypudding/hybrid-ds.

Rongzhao Zhang, Zhian Bai, Ruoying Yu et al.

When delineating lesions from medical images, a human expert can always keep in mind the anatomical structure behind the voxels. However, although high-quality (though not perfect) anatomical information can be retrieved from computed tomography (CT) scans with modern deep learning algorithms, it is still an open problem how these automatically generated organ masks can assist in addressing challenging lesion segmentation tasks, such as the segmentation of colorectal cancer (CRC). In this paper, we develop a novel Anatomy-Guided segmentation framework to exploit the auto-generated organ masks to aid CRC segmentation from CT, namely AG-CRC. First, we obtain multi-organ segmentation (MOS) masks with existing MOS models (e.g., TotalSegmentor) and further derive a more robust organ of interest (OOI) mask that may cover most of the colon-rectum and CRC voxels. Then, we propose an anatomy-guided training patch sampling strategy by optimizing a heuristic gain function that considers both the proximity of important regions (e.g., the tumor or organs of interest) and sample diversity. Third, we design a novel self-supervised learning scheme inspired by the topology of tubular organs like the colon to boost the model performance further. Finally, we employ a masked loss scheme to guide the model to focus solely on the essential learning region. We extensively evaluate the proposed method on two CRC segmentation datasets, where substantial performance improvement (5% to 9% in Dice) is achieved over current state-of-the-art medical image segmentation models, and the ablation studies further evidence the efficacy of every proposed component.

Yankai Jiang, Zhongzhen Huang, Rongzhao Zhang et al.

The long-tailed distribution problem in medical image analysis reflects a high prevalence of common conditions and a low prevalence of rare ones, which poses a significant challenge in developing a unified model capable of identifying rare or novel tumor categories not encountered during training. In this paper, we propose a new zero-shot pan-tumor segmentation framework (ZePT) based on query-disentangling and self-prompting to segment unseen tumor categories beyond the training set. ZePT disentangles the object queries into two subsets and trains them in two stages. Initially, it learns a set of fundamental queries for organ segmentation through an object-aware feature grouping strategy, which gathers organ-level visual features. Subsequently, it refines the other set of advanced queries that focus on the auto-generated visual prompts for unseen tumor segmentation. Moreover, we introduce query-knowledge alignment at the feature level to enhance each query's discriminative representation and generalizability. Extensive experiments on various tumor segmentation tasks demonstrate the performance superiority of ZePT, which surpasses the previous counterparts and evidence the promising ability for zero-shot tumor segmentation in real-world settings.

Yuan Tian, Shuo Wang, Rongzhao Zhang et al.

Medical imaging has significantly advanced computer-aided diagnosis, yet its re-identification (ReID) risks raise critical privacy concerns, calling for de-identification (DeID) techniques. Unfortunately, existing DeID methods neither particularly preserve medical semantics, nor are flexibly adjustable towards different privacy levels. To address these issues, we propose a divide-and-conquer framework comprising two steps: (1) Identity-Blocking, which blocks varying proportions of identity-related regions, to achieve different privacy levels; and (2) Medical-Semantics-Compensation, which leverages pre-trained Medical Foundation Models (MFMs) to extract medical semantic features to compensate the blocked regions. Moreover, recognizing that features from MFMs may still contain residual identity information, we introduce a Minimum Description Length principle-based feature decoupling strategy, to effectively decouple and discard such identity components. Extensive evaluations against existing approaches across seven datasets and three downstream tasks, demonstrates our state-of-the-art performance.

Mouxiao Bian, Rongzhao Zhang, Chao Ding et al.

Large Language Models (LLMs) are poised to transform healthcare under China's Healthy China 2030 initiative, yet they introduce new ethical and patient-safety challenges. We present a novel 12,000-item Q&A benchmark covering 11 ethics and 9 safety dimensions in medical contexts, to quantitatively evaluate these risks. Using this dataset, we assess state-of-the-art Chinese medical LLMs (e.g., Qwen 2.5-32B, DeepSeek), revealing moderate baseline performance (accuracy 42.7% for Qwen 2.5-32B) and significant improvements after fine-tuning on our data (up to 50.8% accuracy). Results show notable gaps in LLM decision-making on ethics and safety scenarios, reflecting insufficient institutional oversight. We then identify systemic governance shortfalls-including the lack of fine-grained ethical audit protocols, slow adaptation by hospital IRBs, and insufficient evaluation tools-that currently hinder safe LLM deployment. Finally, we propose a practical governance framework for healthcare institutions (embedding LLM auditing teams, enacting data ethics guidelines, and implementing safety simulation pipelines) to proactively manage LLM risks. Our study highlights the urgent need for robust LLM governance in Chinese healthcare, aligning AI innovation with patient safety and ethical standards.

Jinru Ding, Lu Lu, Chao Ding et al.

Recent advances in medical large language models (LLMs), multimodal models, and agents demand evaluation frameworks that reflect real clinical workflows and safety constraints. We present MedBench v4, a nationwide, cloud-based benchmarking infrastructure comprising over 700,000 expert-curated tasks spanning 24 primary and 91 secondary specialties, with dedicated tracks for LLMs, multimodal models, and agents. Items undergo multi-stage refinement and multi-round review by clinicians from more than 500 institutions, and open-ended responses are scored by an LLM-as-a-judge calibrated to human ratings. We evaluate 15 frontier models. Base LLMs reach a mean overall score of 54.1/100 (best: Claude Sonnet 4.5, 62.5/100), but safety and ethics remain low (18.4/100). Multimodal models perform worse overall (mean 47.5/100; best: GPT-5, 54.9/100), with solid perception yet weaker cross-modal reasoning. Agents built on the same backbones substantially improve end-to-end performance (mean 79.8/100), with Claude Sonnet 4.5-based agents achieving up to 85.3/100 overall and 88.9/100 on safety tasks. MedBench v4 thus reveals persisting gaps in multimodal reasoning and safety for base models, while showing that governance-aware agentic orchestration can markedly enhance benchmarked clinical readiness without sacrificing capability. By aligning tasks with Chinese clinical guidelines and regulatory priorities, the platform offers a practical reference for hospitals, developers, and policymakers auditing medical AI.

Yuan Tian, Min Zhou, Yitong Chen et al.

Patient face images provide a convenient mean for evaluating eye diseases, while also raising privacy concerns. Here, we introduce ROFI, a deep learning-based privacy protection framework for ophthalmology. Using weakly supervised learning and neural identity translation, ROFI anonymizes facial features while retaining disease features (over 98\% accuracy, $κ> 0.90$). It achieves 100\% diagnostic sensitivity and high agreement ($κ> 0.90$) across eleven eye diseases in three cohorts, anonymizing over 95\% of images. ROFI works with AI systems, maintaining original diagnoses ($κ> 0.80$), and supports secure image reversal (over 98\% similarity), enabling audits and long-term care. These results show ROFI's effectiveness of protecting patient privacy in the digital medicine era.

Zhongzhen Huang, Yankai Jiang, Rongzhao Zhang et al.

Existing promptable segmentation methods in the medical imaging field primarily consider either textual or visual prompts to segment relevant objects, yet they often fall short when addressing anomalies in medical images, like tumors, which may vary greatly in shape, size, and appearance. Recognizing the complexity of medical scenarios and the limitations of textual or visual prompts, we propose a novel dual-prompt schema that leverages the complementary strengths of visual and textual prompts for segmenting various organs and tumors. Specifically, we introduce CAT, an innovative model that Coordinates Anatomical prompts derived from 3D cropped images with Textual prompts enriched by medical domain knowledge. The model architecture adopts a general query-based design, where prompt queries facilitate segmentation queries for mask prediction. To synergize two types of prompts within a unified framework, we implement a ShareRefiner, which refines both segmentation and prompt queries while disentangling the two types of prompts. Trained on a consortium of 10 public CT datasets, CAT demonstrates superior performance in multiple segmentation tasks. Further validation on a specialized in-house dataset reveals the remarkable capacity of segmenting tumors across multiple cancer stages. This approach confirms that coordinating multimodal prompts is a promising avenue for addressing complex scenarios in the medical domain.

Linrui Dai, Rongzhao Zhang, Yongrui Yu et al.

The recently emerging conditional diffusion models seem promising for mitigating the labor and expenses in building large 3D medical imaging datasets. However, previous studies on 3D CT generation have yet to fully capitalize on semantic and textual conditions, and they have primarily focused on specific organs characterized by a local structure and fixed contrast. In this work, we present GuideGen, a controllable framework that generates anatomical masks and corresponding CT volumes for the entire torso-from chest to pelvis-based on free-form text prompts. Our approach includes three core components: a text-conditional semantic synthesizer for creating realistic full-torso anatomies; a contrast-aware autoencoder for detailed, high-fidelity feature extraction across varying contrast levels; and a latent feature generator that ensures alignment between CT images, anatomical semantics and input prompts. To train and evaluate GuideGen, we compile a multi-modality cancer imaging dataset with paired CT and clinical descriptions from 12 public TCIA datasets and one private real-world dataset. Comprehensive evaluations across generation quality, cross-modality alignment, and data usability on multi-organ and tumor segmentation tasks demonstrate GuideGen's superiority over existing CT generation methods.

Rongzhao Zhang, Albert C. S. Chung

When introducing advanced image computing algorithms, e.g., whole-heart segmentation, into clinical practice, a common suspicion is how reliable the automatically computed results are. In fact, it is important to find out the failure cases and identify the misclassified pixels so that they can be excluded or corrected for the subsequent analysis or diagnosis. However, it is not a trivial problem to predict the errors in a segmentation mask when ground truth (usually annotated by experts) is absent. In this work, we attempt to address the pixel-wise error map prediction problem and the per-case mask quality assessment problem using a unified deep learning (DL) framework. Specifically, we first formalize an error map prediction problem, then we convert it to a segmentation problem and build a DL network to tackle it. We also derive a quality indicator (QI) from a predicted error map to measure the overall quality of a segmentation mask. To evaluate the proposed framework, we perform extensive experiments on a public whole-heart segmentation dataset, i.e., MICCAI 2017 MMWHS. By 5-fold cross validation, we obtain an overall Dice score of 0.626 for the error map prediction task, and observe a high Pearson correlation coefficient (PCC) of 0.972 between QI and the actual segmentation accuracy (Acc), as well as a low mean absolute error (MAE) of 0.0048 between them, which evidences the efficacy of our method in both error map prediction and quality assessment.