DRRG: A Discrete Diffusion Framework for Radiology Report Generation

PDD-RRG: Posterior Diagnostic Decision for Study-level Radiology Report Generation

Automatic radiology report generation (RRG) aims to simulate the workflow of radiologists, assisting them in clinical diagnosis. However, existing methods often fall short in utilizing all information relevant to the examination, as is typically done in clinical practice. Although some works attempt to incorporate multi-view images and historical data, these additional inputs may sometimes lead to avoidable diagnostic errors on the contrary. To address these challenges, we introduce a decision-making stage after report generation for the first time and propose a Posterior Diagnostic Decision framework (PDD-RRG) to integrate potentially conflicting diagnoses. Specifically, we create various subsets of input data and utilize an existing RRG model to generate reports from different perspectives. Then the Bayesian posterior probability and the learned thresholds for each clinical observation are calculated to obtain an aggregated diagnostic conclusion, which is subsequently used to refine the generated report. Experiments on MIMIC-CXR demonstrate that our proposed PDD-RRG can effectively enhance the clinical efficacy of existing RRG models without any retraining.

ProFuseGPT: Progressive Fusion with Contrastive Refinement for Long-Sequence Medical Report Generation

Automatic medical report generation (MRG) holds promise for alleviating radiologists’ workload, which has spurred growing interest in MRG for stroke diagnosis. However, existing approaches often fail to effectively model its long-range spatial dependencies and suppress phase-level noise in multi-slice sequences, leading to diluted pathological signals and unstable cross-modal alignment. To address this, we propose a novel framework integrating a progressive fusion mechanism (PFM) and intermediate state refinement via contrastive alignment (ISRCA), inspired by radiologists' clinical workflow. PFM progressively refines pathological representations through anatomically deviation-aware adaptive weighting, suppressing noise from normal slices while enhancing salient abnormalities from local deviations to global context integration. ISRCA adopts a teacher-student distillation approach using contrastive learning to mitigate noise propagation and stabilize intermediate report features. Experimental results on two stroke imaging datasets demonstrate that our method outperforms existing approaches in natural language generation (NLG) metrics, highlighting the effectiveness of PFM and ISRCA in handling long-sequence medical images and advancing stroke imaging report generation.

RADAR: A Difference-Aware Retrieval with Organ-Level Alignment for Change Radiology Report Generation.

Automated report generation is limited to static, single-image analysis, failing to address the critical clinical need for longitudinal comparison in monitoring disease progression and treatment efficacy. To bridge this gap, we introduce the new task of Change Radiology Report Generation (CRRG) which aims to automatically generate a comparative radiology report describing interval changes between a patient's current and prior radiological images. To address this challenge, we propose RADAR, a novel framework that integrates deep visual comparison with knowledge-rich text generation. For robust visual analysis, RADAR employs an "align first, then compare" strategy via an organ-level alignment module, to first mitigate non-pathological artifacts before precisely identifying key pathological changes. Our text generation framework integrates soft prompts for visual evidence with a structured four-step workflow for clinical reasoning. Factual accuracy is enhanced by our Knowledge-Infused Generation (KIG) component, which dynamically retrieves knowledge from a database of similar pathological cases. To anchor our newly proposed task of CRRG and facilitate robust evaluation, we introduce the first comprehensive benchmark, featuring a meticulously curated and processed dataset. Extensive experiments on our benchmark demonstrate that RADAR outperforms existing methods on most evaluation metrics. Our work successfully elevates automated radiological assessment from static analysis to dynamic monitoring, producing interpretable and clinically relevant change reports for longitudinal comparison.

Improved Readability and Translational Instability in LLM-Generated Radiology Reports.

BACKGROUND Large language models (LLMs) show promise for converting complex radiology reports into patient-centric language, but inherent output instability may limit clinical application. OBJECTIVES To quantitatively assess the translational accuracy, error rates, and instability of various LLMs when generating patient-centric radiology reports, and evaluate demographic influences on report readability. MATERIALS AND METHODS This retrospective study evaluated 320 de-identified radiology reports processed by three LLMs using a two-stage (baseline and optimized) prompt engineering strategy. Two senior radiologists evaluated medical accuracy, completeness, and recommendation suitability. Readability was evaluated by 16 non-medical participants stratified by age and education. RESULTS Professional radiological evaluation revealed that all tested models exhibited inherent instability, omitted information, and tended to generate risk-averse, generalized clinical recommendations. To address these limitations, optimized structured prompts significantly reduced model output variance and improved translational accuracy, with particularly prominent effects observed in DeepSeek-R1 and ChatGPT-4.0. Overall, large language models significantly enhanced the readability of radiology reports (P < 0.05), with DeepSeek-R1 achieving the best performance. However, patients' self-reported comprehension of the reports was affected by demographic characteristics. CONCLUSION Large language models can effectively improve the readability of radiology reports, yet all such models inherently suffer from output instability and information omission. Optimized structured prompting can substantially reduce the variability of model outputs and improve the accuracy of medical text translation. Nevertheless, LLMs should currently be strictly confined to human-supervised auxiliary tools rather than applied as standalone clinical solutions.

SCAN-R: bridging precision imaging and natural language for automated stroke diagnosis

Stroke is the second leading cause of death globally, where each minute of treatment delay results in the loss of 1.9 million brain cells. Traditional CT-based diagnosis relies on manual interpretation with inherent variability and time constraints, while existing AI approaches typically address only isolated tasks such as detection or segmentation without integrated clinical reporting. We present Stroke CT Analysis and Natural Language Reporting (SCAN-R), a unified end-to-end framework that integrates multiclass stroke detection, Transformerenhanced U-Net segmentation with task-specific pre-trained backbones, and Retrieval-Augmented Generation for evidence-based clinical report generation. Evaluation on 6,653 CT scans demonstrates 95.81% detection accuracy and a Dice coefficient of 0.81 for bleeding lesion segmentation, and 10% improvement in clinical decision-making quality on the MedMCQA benchmark. The framework successfully transforms raw CT images into structured clinical reports with quantitative metadata and evidence-based recommendations, demonstrating potential to accelerate time-critical stroke diagnosis in emergency settings.

Boltzmann-driven dynamic annealing network for knowledge-guided radiology report generation