Papers for

radiology it teams

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

AgentCT improves CT scan restoration by choosing specific fixes

AgentCT: A Vision-Language Agent for Universal CT Imaging

Abstract: Computed tomography (CT) restoration methods often use a single network for low-dose CT (LDCT), sparse-view CT (SVCT), and metal artifact reduction (MAR), despite the distinct physics of these degradations. We propose AgentCT, a vision-language agent that identifies the degradation and routes images to task-specific restoration models. A LoRA-adapted Qwen3-VL-4B controller predicts degradation type and severity, selects a specialist, and reassesses the restored image. The LDCT specialist uses an image-domain diffusion model, whereas the SVCT and MAR specialists combine projection-domain correction, data consistency, differentiable filtered back-projection, and image refinement. We evaluated the controller and specialists separately on patient-disjoint Mayo 2020 abdominal CT data across three severity levels for each degradation. The controller achieved 95.7% accuracy for degradation type and 93.9% for severity, with a 100% JSON parse rate. Using ground-truth degradation labels for routing, AgentCT achieved an overall PSNR/SSIM of 46.18 dB/0.984, compared with 44.93 dB/0.976 for AMIR and 44.29 dB/0.979 for TAMP, and attained the highest PSNR in all nine settings. For SVCT, its dual-domain specialist achieved 46.04 dB with 4.2 million parameters, compared with 44.43 dB for single-task AMIR and 43.57 dB for all-in-one AMIR with 23.5 million parameters. These results support degradation-specific CT restoration with controller-based model selection; evaluation under automatic routing and on independent clinical data remains necessary. Code is available at https://github.com/hao1635/AgentCT.

Sun 8 NovComputer Vision and Pattern Recognition
The gist
CT scans can have different kinds of problems like being too noisy, having missing views, or metal causing distortions. The paper presents AgentCT, a system that first figures out what kind of problem a CT image has and then sends it to a specialized fix designed for that problem. AgentCT’s controller can identify the problem type and severity accurately and its approach leads to clearer CT images compared to other methods. This helps doctors get better images without needing one general fixer that may not work well for all problems.
Open → 2611.11503v1

Multimodal fusion boosts chest X-ray diagnosis accuracy

Guided Co-Attention with Adaptive Gating for Multimodal Medical Data Fusion

Abstract: Multimodal fusion requires not only exchanging information across modalities but also determining which representations should participate in the interaction and how complementary cross-modal views should be combined. We introduce Guided Co-Attention with Adaptive Gating, a fusion framework that integrates token-wise guidance, bidirectional cross-attention, and sample-specific fusion. The proposed method constructs two directional representations: an image-conditioned text representation, obtained by allowing textual tokens to query visual features, and a text-conditioned image representation formed in the reverse direction. Token-wise guidance modulates the representations before cross-attention, while an adaptive gate learns the relative contribution of the two resulting directional representations for each sample. We evaluate the approach using paired chest radiographs and radiology reports as an image-text test bed for binary (normal vs. abnormal) classification. The two cross-attention directions exhibit distinct performance characteristics, demonstrating that reversing the query and source modalities does not produce interchangeable representations. Ablation experiments further show that bidirectional interaction, adaptive gating, and token guidance provide complementary benefits. The proposed model achieves 96.05% accuracy, 96.10% F1-score, 98.37% AUC, 95.60% Sens@95Spec, and 96.51% Spec@95Sens, yielding the highest mean performance across all evaluated metrics. These findings support explicitly modeling interaction direction, token relevance, and sample-dependent integration when constructing multimodal representations.

Sat 7 NovComputer Vision and Pattern RecognitionMachine Learning
The gist
Combining information from different sources like images and text can help make better medical decisions. This paper shows a way to merge chest X-rays and their reports so that both types of information help each other smartly. The method treats image and text differently and decides how much each should influence the result for every case. This approach led to very accurate detection of normal versus abnormal findings in chest X-rays. The authors found that considering both directions of interaction and adjusting the mix per sample improves performance.
Open → 2611.10348v1

Lightweight modules improve medical lesion segmentation accuracy and flexibility

Enhancing Medical Lesion Segmentation via Several Simple Modules

Abstract: This work introduces three lightweight enhancement modules for medical lesion segmentation to improve clinical applicability, particularly under data-scarce and computationally constrained environments. Current mainstream methods struggle with the lack of high-quality annotations for rare pathologies, which often leads to severe morphological distortions during inference, despite achieving seemingly ideal DSC scores. Furthermore, the isolated development of individual segmentation tasks restricts cross-domain methodological innovation, underscoring the need for leveraging shared knowledge across different lesions within the same anatomical region. However, constrained access to historical data and annotations hinders efficient knowledge extraction and triggers catastrophic forgetting when adapting to new classes. To address these challenges, we first propose DT-Morph Combined Loss with continuous spatial pooling to optimize differentiable topological structures and boundary continuity. Second, we introduce 3D Group Prompt, which extracts domain-universal knowledge via curated volumetric image-mask alignment prompts together with text prompts built on SAT baseline. Finally, we replace conventional output layers with Many-to-One Conditional Segmentation Head equipped with synonym-resolved routing, which brings negligible computational cost and enables fast incremental learning for new categories without sacrificing performance on old tasks. Ablation studies on abdominal and extra-abdominal datasets covering both organ and lesion segmentation verify the efficacy of our modules.

Fri 30 OctComputer Vision and Pattern RecognitionMachine Learning
The gist
Medical image analysis often struggles to correctly outline lesions when there isn’t enough quality data or computing power. The authors developed three simple improvements that help software better understand the shapes and boundaries of lesions and remember what it learned about different lesion types. These changes also let the system learn new lesion categories quickly without forgetting old ones. Tests on different medical image sets showed that these changes improve lesion and organ segmentation.
Open → 2610.41652v1

Voxel embedding improves one-shot landmark localization in volume scans

UNIVERS: Learning UNIversal Voxel-wise Embedding via Randomized Synthesis for generalizable one-shot landmark localization

Abstract: One-shot landmark localization in volume scans with varying fields of view (FOVs) is challenging. While existing self-supervised learning methods have shown promising performance, they rely on contrastive learning built upon augmented views of the same image, which inherently limits their generalization in cross-modality scenarios. This motivates research on cross-modality embedding learning using paired multi-modality images. However, existing methods suffer from not only limited training data, but also the requirement of multi-stage iterative training. To address these limitations, we propose UNIVERS, an end-to-end trainable framework that learns UNIversal Voxel-wise Embedding via Randomized Synthesis. Our framework leverages diverse pseudo-data synthesized from organ shape templates, with randomization applied to semantic layout, visual appearance, FOV, and spatial pose. Coupled with cross-domain discriminative voxel-wise contrastive learning, this domain randomization strategy forces the network to disregard domain-specific biases and learn features that are robust against variations in imaging contrast, resolution, and spatial context. Consequently, the trained model enables one-shot landmark localization in multi-modality volume scans with arbitrary FOVs. In this study, the efficacy of the proposed method is demonstrated by applying it to the task of localizing the adrenal glands, which are paired, small, and irregular endocrine organs. In particular, comprehensive experiments are conducted on three challenging adrenal gland datasets. The experimental results demonstrate that UNIVERS consistently outperforms the state-of-the-art (SOTA) one-shot landmark localization approaches, with particularly prominent advantages in cross-modality scenarios. The source code is publicly available at https://github.com/Kendra9999/UNIVERS.

Wed 21 OctComputer Vision and Pattern RecognitionArtificial Intelligence
The gist
Finding specific landmarks within 3D medical images is hard, especially when images come from different types of scans or cover different body areas. The researchers created a new technique called UNIVERS that trains a computer model using many synthetic images with varied features, helping it recognize landmarks more reliably across different scan types. This approach performs better than existing methods when identifying small organs like adrenal glands in diverse medical images. UNIVERS works in just one shot, meaning it can find landmarks using only one example image for reference.
Open → 2610.29098v1

Class conditioned explanation guidance improves medical image analysis accuracy

Learning from the Right Regions: Class-Conditioned Explanation Guidance for Medical Imaging

Abstract: Deep neural networks can achieve strong predictive performance in medical imaging while relying on regions unrelated to the predicted finding. Explanation-guided learning can address this problem, but multi-label classification introduces a key challenge: multiple abnormalities may coexist in an image, requiring each prediction to align with its corresponding spatial evidence. We propose a class-conditioned explanation-guided learning framework that constructs a separate explanation for each positive class and penalizes evidence outside its class-specific annotated regions. We instantiate the framework using Gradient-weighted Class Activation Mapping (Grad-CAM) and evaluate it on 14 abnormalities using EfficientNet-B1, ViT-B/16, ViT-B/32, and Swin-B. Explanation guidance substantially improves annotation alignment across all architectures: All-Saliency Precision increases from 0.140 to 0.625, 0.052 to 0.551, 0.060 to 0.403, and 0.095 to 0.570, respectively. These improvements are accompanied by smaller, architecture-dependent changes in classification performance. Our results show that class-conditioned explanation guidance can substantially improve where multi-label classifiers concentrate their explanatory evidence while largely preserving predictive performance.

Sat 17 OctComputer Vision and Pattern RecognitionMachine Learning
The gist
Sometimes AI models that analyze medical images make predictions based on the wrong parts of the image, which can be confusing or misleading. The authors developed a new approach that lets the AI focus on the right areas linked to each specific condition detected in an image, even when multiple conditions appear together. They tested their method with different AI systems and found the explanations became much better aligned with the correct regions, while the model’s prediction accuracy stayed about the same. This method helps make AI decisions in medical images more understandable and trustworthy.
Open → 2610.24381v1

Medical vision language models retain key image details after token pruning

MedSTP: Preserving Subspace and Transition Evidence for Medical VQA Token Pruning

Abstract: Medical vision-language models (VLMs) for visual question answering (VQA) often pass long visual-token sequences to the language model, making visual-token pruning attractive but risky for medical images, where answers may depend on localized boundary and texture evidence. We study training free pre-decoding pruning and treat pruning as a subspace stable evidence preservation problem. The retained tokens should remain close to the dominant visual evidence subspace induced by the full token set. To this end, we propose \emph{Subspace-Transition Token Pruning} (\method{}), which estimates each token's spectral support from principal component space and constructs a compact representation subspace. To better reflect medical image structure, \method{} further uses a bounded token-grid transition prior to promote boundary and texture sensitive tokens. We evaluate \method{} on the closed-ended SLAKE and OmniMedVQA benchmarks and on RadImageNet-VQA, which provides both closed-ended and open-ended evaluations, using Lingshu-7B and Qwen3-VL-8B backbones. \method{} delivers consistently strong performance on both the closed-ended and open-ended RadImageNet-VQA evaluations. These results show that \method{} is an effective training free pruning framework for medical domain VQA.

Thu 15 OctComputer Vision and Pattern Recognition
The gist
Medical vision-language models answer questions about medical images using many small image pieces called tokens. The authors found a way to remove some tokens without losing important details like edges and textures that help answer questions accurately. They developed a method called MedSTP that keeps tokens representing the main visual features and those near boundaries. This improves efficiency while still performing well in tests on medical image question answering tasks.
Open → 2610.22751v1

Routine CT report markings improve 3D image finding localization

Scaling 3D Visual Grounding in Abdominal CT

Abstract: Visual grounding models can enhance radiology workflows by linking report findings to image regions. This is particularly valuable for 3D CT, where findings often occupy a tiny fraction of the volume. Training 3D grounding models requires large sets of paired phrases and regions, and building such datasets is expensive, requiring radiologists to annotate images by hand. We posit that this supervision is already created implicitly during routine reporting, as radiologists frequently place 2D annotations (e.g., distance measurement, arrows) on key images to make measurements and to support report interpretation. We introduce an automated pipeline that converts these routine clinical annotations into large-scale phrase-region supervision for 3D visual grounding. The pipeline links each annotation to the corresponding finding in the report through metadata matching, then uses a promptable 3D segmentation model to convert the 2D annotation into a volumetric mask. This produces phrase-mask-volume datasets without requiring additional radiologist annotation. Applied to a single institution's clinical picture archiving and communication system (PACS), our approach generated 105K phrase-mask-volume triplets from 59K abdominal CT exams. We also introduce two abdominal CT grounding benchmarks, LocusBench-Onc and LocusBench-ED, which comprise 240 oncology and 260 emergency-department radiologist-reviewed phrase-mask-volume triplets, respectively, with the latter spanning 13 distinct categories such as appendicitis, hematoma, and hernia. We further introduce LocusCT, a 3D visual grounding model trained on this dataset, which achieves hit rates of 0.725 on LocusBench-Onc and 0.773 on LocusBench-ED, substantially outperforming comparator models. These results show that routine PACS annotations are a scalable, previously unused source of supervision for 3D visual grounding.

Fri 2 OctComputer Vision and Pattern RecognitionMachine Learning
The gist
Doctors often add markings like arrows and measurements to CT images when writing reports, but these annotations are typically only seen in 2D slices. The authors show how to automatically convert these 2D annotations into 3D regions within abdominal CT scans and link them to the corresponding text in reports. This creates a large dataset without extra work from doctors, which can train models to better find and highlight important features in 3D scans. Their model performs well on two new benchmarks involving different medical conditions in CT images.
Open → 2610.04095v1

KneePreM improves knee MRI analysis with less labeled data

KneePreM: Towards 3D Knee MRI Foundation Models via Large-Scale Unlabeled Pretraining and Label-Efficient Fine-Tuning

Abstract: Background: Large volumes of unlabeled knee MRI scans are available across repositories but remain insufficiently leveraged. We developed KneePreM, a knee-specific 3D self-supervised model, and evaluated transfer and label efficiency for classification and segmentation. Methods: A 3D U-Net masked autoencoder was pretrained on 19,011 unlabeled Osteoarthritis Initiative (OAI) MRI series from 4,791 participants. Downstream fine-tuning used full and reduced training sets for fastMRI+ two-label classification (1,172 examinations), Arthroscopic Partial Meniscectomy (APM) eight-target classification (1,716 examinations), SKM-TEA segmentation (155 examinations), and APM segmentation (25 examinations). Baselines were random initialization and SuPreM. Deployment workflow was implemented with a Model Context Protocol interface. Evaluation metrics included balanced accuracy, F1 score, ROC AUC, PR AUC, and Dice score. Statistical analysis used bootstrap confidence intervals and paired bootstrap tests for classification and Wilcoxon signed-rank tests for segmentation. Results: KneePreM achieved higher full-data macro ROC AUC than both baselines for fastMRI+ and APM (all p < .001). For fastMRI+ classification, KneePreM achieved a ROC AUC of 0.722 using 50% of the training data, exceeding both full-data baselines. In APM classification, KneePreM reached a ROC AUC of 0.740 with 70% of the data, matching the full-data random baseline and outperforming SuPreM. For SKM-TEA segmentation, its 70%-data Dice of 0.838 exceeded the full-data random baseline (0.835) and both same-budget comparators. In APM segmentation, its 75%-data Dice of 0.746 exceeded the full-data random baseline (0.731) and both same-budget comparators. Conclusion: KneePreM improves transfer performance and label efficiency across knee MRI classification and segmentation tasks, particularly when labeled training data are limited.

Fri 25 SeptComputer Vision and Pattern Recognition
The gist
Analyzing knee MRI scans can be difficult because labeled examples are scarce even though many unlabeled scans exist. The authors created KneePreM, a computer model trained on many unlabeled knee MRIs that can then learn new tasks with fewer labeled images. This approach improved the accuracy of knee condition classification and tissue segmentation, especially when less labeled data was available. Their method outperformed previous models on several knee MRI tasks.
Open → 2609.31461v1

Generalist specialist AI improves rare disease detection in medical images

Generalist-Specialist Mixture-of-Experts for Rare Pathology Detection in Multimodal Imaging

Abstract: AI models for multimodal medical imaging must balance modality-specific specialization with cross-modal shared representations, a trade-off that pure Mixture-of-Experts (MoE) architectures currently fail to satisfy. Expert-based routing improves in-domain learning but may sacrifice cross-modal signals, which appear particularly important for rare (low-prevalence) pathologies in our experiments. To resolve this, we introduce Generalist-Specialist-MoE (GS-MoE), a two-branch (MoE) architecture that couples a cross-modal generalist model with distinct modality-specific specialists (experts) via domain-constrained feature fusion. On RadImageNet (1.35M images, 165 pathologies, three modalities), GS-MoE recovers detection of six low-prevalence pathologies on which every baseline scores F1 $=$ 0, with per-class gains up to +0.60 F1. It attains this while even slightly exceeding dense and specialist-only MoE aggregate baselines (MCC 0.770), while using ${\sim}53\%$ fewer active parameters at inference than the strongest investigated dense model.

Wed 16 SeptComputer Vision and Pattern RecognitionArtificial Intelligence
The gist
Detecting rare diseases from medical images is hard because each imaging type needs special knowledge, but sharing information across types can help. The paper shows that usual AI models with multiple experts don't balance this well. The authors propose a new model that uses both general knowledge across imaging types and specific knowledge for each type, working together. This approach better identifies rare diseases and uses fewer computing resources.
Open → 2609.18688v1

Medical report accuracy improves with concept grounded reasoning and lesion localization

Concept-Grounded Reasoning with Prompt-Driven Localization for Interpretable Structured Report Generation

Abstract: Medical imaging modalities such as ultrasound and X-ray are widely used in clinical practice, where diagnosis follows a structured, evidence-driven workflow aligned with standardized criteria. While multimodal large language models (MLLMs) show promise for automated medical report generation, most existing systems rely on end-to-end multimodal fusion without modeling clinically defined intermediate attributes, leading to limited grounding and interpretability. To address this issue, we propose CORAL (COncept-grounded ReAsoning with Localization), a multimodal framework that integrates spatial grounding and concept-level supervision into a unified reasoning process. CORAL employs a prompt-driven medical segmentation model to localize lesions and predicts multi-class clinical attributes through a Concept Bottleneck module. The resulting textual concept tokens are combined with mask-modulated visual features within an MLLM to enable structured report generation and diagnostic prediction. Experiments on BUS-CoT and IU X-ray datasets demonstrate consistent improvements in diagnostic accuracy, concept consistency, and report quality over strong general-purpose and medical MLLMs, indicating that concept-grounded reasoning better aligns generation with clinical decision processes.

Mon 14 SeptComputer Vision and Pattern RecognitionArtificial IntelligenceMachine Learning
The gist
Medical images like ultrasounds and X-rays are used by doctors to make detailed reports based on what they see in the images. Current computer systems that create these reports often miss important clinical details and are hard to understand. The authors developed a new system called CORAL that finds specific problem areas in the images and uses medical concepts to guide the report generation. This approach helps the system align better with how doctors think and improves the accuracy and quality of the reports.
Open → 2609.15334v1