Text Over Image: Auditing Multimodal Robustness in Synthetic Medical Image Detection

• URL: http://arxiv.org/abs/2606.25375v2
• Date: Wed, 01 Jul 2026 07:46:36 GMT
• Title: Text Over Image: Auditing Multimodal Robustness in Synthetic Medical Image Detection
• Abstract summary: We reformulate synthetic medical image detection as an audit of multimodal robustness at the image-record interface.<n>We find that changing the metadata context alone can flip authenticity judgments.<n>We propose an inference-time mitigation pipeline that detects and neutralizes provenance shortcuts without model retraining.
• Score: 38.146353416394234
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: With the rapid adoption of generative AI, synthetic medical images pose growing risks, including diagnostic deception and insurance fraud. Although prior work has explored vision-language model (VLM)-based synthetic image detection, these evaluations typically consider images in isolation. In clinical practice, however, images are interpreted alongside structured records and metadata, and VLMs are increasingly deployed under joint image-record inputs. We uncover a previously underexamined multimodal vulnerability: when given both modalities, VLMs may overweight record context in authenticity judgments, such that the same image receives different predictions solely due to changes in its accompanying text. This raises concerns about robustness in real-world deployment. To systematically characterize this effect, we reformulate synthetic medical image detection as an audit of multimodal robustness at the image-record interface and introduce a paired benchmark that holds the image fixed while swapping controlled metadata variants. Across multiple imaging modalities, we evaluate diverse open-weight and frontier API VLMs and find that changing the metadata context alone can flip authenticity judgments, with accuracy on authentic images dropping by 61.1% on average under an explicit AI-origin tag. We further propose an inference-time mitigation pipeline that detects and neutralizes provenance shortcuts without model retraining, substantially outperforming direct prompt-based suppression on the affected subset. Our benchmark provides a standardized tool for assessing and improving multimodal robustness beyond image-only settings. Code and data will be released upon acceptance.

Related papers

• Beyond Static Anchors: Bounded Prototype Conditioning for Language-Free Medical Anomaly Detection [52.986874008247554]
  We propose ReCAP, a language-free framework that replaces static anchors with input-conditioned visual prototypes.<n>ReCAP re-centers separated normal and abnormal prototypes for each image through a bounded gated modulation.<n>It achieves the best image-level AUROC on all zero-shot and 23 of 24 few-shot settings, and the best zero-shot pixel-level AUROC on all three segmentation datasets.
  arXiv  Detail & Related papers  (2026-08-01T04:46:55Z)
• Case-Aware Medical Image Classification with Multimodal Knowledge Graphs and Reliability-Guided Refinement [27.443476988184994]
  We propose a case-aware reasoning framework driven by multimodal knowledge graphs for medical image classification.<n>Given an input image, our method adaptively retrieves similar cases from this memory and extracts their corresponding case-centered subgraphs.<n>Experiments on multiple medical imaging datasets demonstrate that our approach consistently outperforms strong baselines.
  arXiv  Detail & Related papers  (2026-05-21T14:29:20Z)
• GenShield: Unified Detection and Artifact Correction for AI-Generated Images [65.11434977803509]
  GenShield is a framework that performs AIGI detection and artifact correction in a closed loop from diagnosis to restoration.<n>A high-quality dataset with large-scale artifact-restored'' pairs is constructed alongside a unified evaluation pipeline.
  arXiv  Detail & Related papers  (2026-05-15T16:06:20Z)
• More Images, More Problems? A Controlled Analysis of VLM Failure Modes [80.64323947730905]
  Large Vision Language Models (LVLMs) have demonstrated remarkable capabilities, yet their proficiency in understanding and reasoning over multiple images remains largely unexplored.<n>We introduce MIMIC, a new benchmark designed to rigorously evaluate the multi-image capabilities of LVLMs.
  arXiv  Detail & Related papers  (2026-01-12T18:45:13Z)
• INSIGHT: An Interpretable Neural Vision-Language Framework for Reasoning of Generative Artifacts [0.0]
  Current forensic systems degrade sharply under real-world conditions.<n>Most detectors operate as opaques, offering little insight into why an image is flagged as synthetic.<n>We introduce INSIGHT, a unified framework for robust detection and transparent explanation of AI-generated images.
  arXiv  Detail & Related papers  (2025-11-27T11:43:50Z)
• From Evidence to Verdict: An Agent-Based Forensic Framework for AI-Generated Image Detection [19.240335260177382]
  We introduce AIFo (Agent-based Image Forensics), a training-free framework that emulates human forensic investigation through multi-agent collaboration.<n>Unlike conventional methods, our framework employs a set of forensic tools, including reverse image search, metadata extraction, pre-trained classifiers, and VLM analysis.<n>Our comprehensive evaluation spans 6,000 images and challenges real-world scenarios, including images from modern generative platforms and diverse online sources.
  arXiv  Detail & Related papers  (2025-10-31T18:36:49Z)
• UltraAD: Fine-Grained Ultrasound Anomaly Classification via Few-Shot CLIP Adaptation [39.48115172323913]
  We propose UltraAD, a vision-language model (VLM)-based approach for anomaly localization and fine-grained classification.<n>UltraAD has been extensively evaluated on three breast US datasets, outperforming state-of-the-art methods in both lesion datasets and fine-grained medical classification.
  arXiv  Detail & Related papers  (2025-06-24T15:00:38Z)
• Explainable Synthetic Image Detection through Diffusion Timestep Ensembling [30.298198387824275]
  We propose a novel synthetic image detection method that directly utilizes features of intermediately noised images by training an ensemble on multiple noised timesteps.<n>To enhance human comprehension, we introduce a metric-grounded explanation generation and refinement module.<n>Our method achieves state-of-the-art performance with 98.91% and 95.89% detection accuracy on regular and challenging samples respectively.
  arXiv  Detail & Related papers  (2025-03-08T13:04:20Z)
• Fréchet Radiomic Distance (FRD): A Versatile Metric for Comparing Medical Imaging Datasets [13.737058479403311]
  We introduce a new perceptual metric tailored for medical images, FRD (Fr'echet Radiomic Distance)<n>We show that FRD is superior to other image distribution metrics for a range of medical imaging applications.<n> FRD offers additional benefits such as stability and computational efficiency at low sample sizes.
  arXiv  Detail & Related papers  (2024-12-02T13:49:14Z)
• Adapting Visual-Language Models for Generalizable Anomaly Detection in Medical Images [68.42215385041114]
  This paper introduces a novel lightweight multi-level adaptation and comparison framework to repurpose the CLIP model for medical anomaly detection. Our approach integrates multiple residual adapters into the pre-trained visual encoder, enabling a stepwise enhancement of visual features across different levels. Our experiments on medical anomaly detection benchmarks demonstrate that our method significantly surpasses current state-of-the-art models.
  arXiv  Detail & Related papers  (2024-03-19T09:28:19Z)
• Improving Adversarial Robustness of Masked Autoencoders via Test-time Frequency-domain Prompting [133.55037976429088]
  We investigate the adversarial robustness of vision transformers equipped with BERT pretraining (e.g., BEiT, MAE) A surprising observation is that MAE has significantly worse adversarial robustness than other BERT pretraining methods. We propose a simple yet effective way to boost the adversarial robustness of MAE.
  arXiv  Detail & Related papers  (2023-08-20T16:27:17Z)
• Learning to Exploit Temporal Structure for Biomedical Vision-Language Processing [53.89917396428747]
  Self-supervised learning in vision-language processing exploits semantic alignment between imaging and text modalities. We explicitly account for prior images and reports when available during both training and fine-tuning. Our approach, named BioViL-T, uses a CNN-Transformer hybrid multi-image encoder trained jointly with a text model.
  arXiv  Detail & Related papers  (2023-01-11T16:35:33Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.