Selected work

Representative Papers

PathLang: A Language-Centered Benchmark for Vision-Language Models in Computational Pathology

Oct 08, 2026

This study addresses the insufficient robustness of pathology vision-language models to variations in diagnostic phrasing. For the first time, language is treated as an independent variable axis to construct a language-centric zero-shot benchmark. By fixing images and labels while systematically perturbing terminology, specificity, and reporting style, this work evaluates models across classification, retrieval, and open-vocabulary diagnosis tasks, with prompts validated by six board-certified pathologists. The findings reveal that model performance is highly sensitive to clinically equivalent paraphrasing and that image-text alignment quality does not necessarily translate into inter-class separability. To facilitate future research, the complete evaluation resources are publicly released.

0 citationsRead paper

SlideBank: A Persistent Hierarchical Evidence Bank for Consistent Whole-Slide Reasoning

Aug 31, 2026

为解决WSI中稀疏、异质和跨尺度的诊断相关形态问题，提出SlideBank框架，通过构建持久性证据库并结合多级证据整合方法进行有效推理。

0 citationsRead paper

Rethinking Artificial Intelligence in Medical Imaging: Assumptions, Reality, and Reframing

Jul 29, 2026

This study addresses the structural misalignment between medical imaging AI research and clinical practice, which impedes the translation of algorithmic performance into tangible diagnostic and therapeutic value. The authors systematically identify six dimensions of misalignment across design, evaluation, and decision-making phases and propose a novel paradigm centered on clinicians, emphasizing actionability and multimodal integration. By incorporating multimodal modeling, interpretable interactive design, few-shot domain adaptation, and seamless integration into clinical workflows, the framework aims to develop “agentive, clinician-aligned” AI systems that generate actionable clinical guidance rather than mere predictions. This work offers the first structural analysis of implementation bottlenecks in medical AI and provides a systematic pathway to bridge the gap between artificial intelligence and real-world clinical practice, thereby enabling AI to genuinely support clinical judgment.

0 citationsRead paper

UltraVR: A Diagnostic Ultra-Resolution Image-VQA Benchmark for Evidence-Grounded Reasoning

Jun 03, 2026

This work addresses the limited evidence-driven reasoning capabilities of current vision-language models on super-resolution images and the absence of fine-grained evaluation protocols for their reasoning processes. To this end, the authors propose UltraVR, a diagnostic visual question answering benchmark tailored to four domains: surveillance, remote sensing, histopathology, and industrial inspection. UltraVR introduces, for the first time, a structured chain-of-thought annotation scheme comprising step-level questions, intermediate answers, and reasoning labels, enabling diagnostic analysis across the full pipeline—from evidence localization and local perception to final decision-making. Experimental results reveal that state-of-the-art models perform poorly in super-resolution reasoning, with errors predominantly occurring in the initial two stages; notably, downstream reasoning accuracy improves substantially when intermediate visual facts are explicitly provided.

0 citationsRead paper

Radiuma: A Unified Zero-Code Executable Graphical Workflow Generator for Reproducible and Shareable Medical Image Analysis and Machine Learning

May 22, 2026

This work addresses the lack of standardized, user-friendly, and reproducible software environments in medical image analysis, which hinders the widespread adoption of advanced methodologies. To overcome this limitation, the authors propose a modular, zero-code platform that enables seamless integration of image reading, visualization, registration, segmentation, radiomics feature extraction, and machine learning through intuitive graphical workflows. The platform allows users to construct, execute, and share end-to-end analysis pipelines without programming expertise. It introduces a unified, executable, and shareable cross-modality workflow system that ensures full transparency and facilitates collaborative reuse across disciplines. Supporting classification, regression, and clustering tasks, the platform significantly enhances analytical consistency, reproducibility, and interdisciplinary collaboration in clinical and translational research, as demonstrated by experimental validation.

0 citationsRead paper

Recent publications

Latest Papers

PathLang: A Language-Centered Benchmark for Vision-Language Models in Computational Pathology

Oct 08, 2026

This study addresses the insufficient robustness of pathology vision-language models to variations in diagnostic phrasing. For the first time, language is treated as an independent variable axis to construct a language-centric zero-shot benchmark. By fixing images and labels while systematically perturbing terminology, specificity, and reporting style, this work evaluates models across classification, retrieval, and open-vocabulary diagnosis tasks, with prompts validated by six board-certified pathologists. The findings reveal that model performance is highly sensitive to clinically equivalent paraphrasing and that image-text alignment quality does not necessarily translate into inter-class separability. To facilitate future research, the complete evaluation resources are publicly released.

0 citationsRead paper

SlideBank: A Persistent Hierarchical Evidence Bank for Consistent Whole-Slide Reasoning

Aug 31, 2026

为解决WSI中稀疏、异质和跨尺度的诊断相关形态问题，提出SlideBank框架，通过构建持久性证据库并结合多级证据整合方法进行有效推理。

0 citationsRead paper

Rethinking Artificial Intelligence in Medical Imaging: Assumptions, Reality, and Reframing

Jul 29, 2026

This study addresses the structural misalignment between medical imaging AI research and clinical practice, which impedes the translation of algorithmic performance into tangible diagnostic and therapeutic value. The authors systematically identify six dimensions of misalignment across design, evaluation, and decision-making phases and propose a novel paradigm centered on clinicians, emphasizing actionability and multimodal integration. By incorporating multimodal modeling, interpretable interactive design, few-shot domain adaptation, and seamless integration into clinical workflows, the framework aims to develop “agentive, clinician-aligned” AI systems that generate actionable clinical guidance rather than mere predictions. This work offers the first structural analysis of implementation bottlenecks in medical AI and provides a systematic pathway to bridge the gap between artificial intelligence and real-world clinical practice, thereby enabling AI to genuinely support clinical judgment.

0 citationsRead paper

UltraVR: A Diagnostic Ultra-Resolution Image-VQA Benchmark for Evidence-Grounded Reasoning

Jun 03, 2026

This work addresses the limited evidence-driven reasoning capabilities of current vision-language models on super-resolution images and the absence of fine-grained evaluation protocols for their reasoning processes. To this end, the authors propose UltraVR, a diagnostic visual question answering benchmark tailored to four domains: surveillance, remote sensing, histopathology, and industrial inspection. UltraVR introduces, for the first time, a structured chain-of-thought annotation scheme comprising step-level questions, intermediate answers, and reasoning labels, enabling diagnostic analysis across the full pipeline—from evidence localization and local perception to final decision-making. Experimental results reveal that state-of-the-art models perform poorly in super-resolution reasoning, with errors predominantly occurring in the initial two stages; notably, downstream reasoning accuracy improves substantially when intermediate visual facts are explicitly provided.

0 citationsRead paper

Radiuma: A Unified Zero-Code Executable Graphical Workflow Generator for Reproducible and Shareable Medical Image Analysis and Machine Learning

May 22, 2026

This work addresses the lack of standardized, user-friendly, and reproducible software environments in medical image analysis, which hinders the widespread adoption of advanced methodologies. To overcome this limitation, the authors propose a modular, zero-code platform that enables seamless integration of image reading, visualization, registration, segmentation, radiomics feature extraction, and machine learning through intuitive graphical workflows. The platform allows users to construct, execute, and share end-to-end analysis pipelines without programming expertise. It introduces a unified, executable, and shareable cross-modality workflow system that ensures full transparency and facilitates collaborative reuse across disciplines. Supporting classification, regression, and clustering tasks, the platform significantly enhances analytical consistency, reproducibility, and interdisciplinary collaboration in clinical and translational research, as demonstrated by experimental validation.

0 citationsRead paper