Scene Graph Aided Radiology Report Generation Study
The author proposes a novel framework, SGRRG, that leverages radiology scene graphs to distill valuable knowledge within each sample for accurate report generation. By integrating scene graph generation processes and abnormal information learning modules, SGRRG outperforms previous methods in capturing abnormal findings.
Large Model-driven Radiology Report Generation with Clinical Quality Reinforcement Learning
The author introduces a novel method, LM-RRG, that combines large models with clinical quality reinforcement learning to generate accurate and comprehensive chest X-ray radiology reports. The approach aims to address the shortcomings of current methods by emphasizing specific regions with medical significance and incorporating a novel clinical quality reinforcement learning strategy.
Memory-based Cross-modal Semantic Alignment Network for Accurate and Fluent Radiology Report Generation
A memory-based cross-modal semantic alignment network is proposed to generate accurate and fluent radiology reports by learning disease-related representations and prior knowledge shared between radiology images and reports, and performing fine-grained feature consolidation with semantic alignment.
Comprehensive Review of Deep Learning-based Radiology Report Generation
Radiology report generation (RRG) aims to automatically generate free-text descriptions from clinical radiographs, which plays an essential role in promoting clinical automation and assisting inexperienced doctors and radiologists.
SERPENT-VLM: A Self-Refining Approach for Accurate and Hallucination-Free Radiology Report Generation Using Vision-Language Models
SERPENT-VLM introduces a self-refining mechanism that leverages the similarity between the pooled image representation and the contextual representation of the generated radiological text to improve the accuracy and coherence of radiology report generation, reducing hallucination.
Improving Inter-Report Consistency in Radiology Report Generation via Lesion-aware Mixup Augmentation
Improving the inter-report consistency of radiology report generation by extracting lesions, examining their characteristics, and using a lesion-aware mixup technique to align the representations of semantically equivalent lesions.
Sparse Autoencoders for Interpretable Radiology Report Generation: Achieving Competitive Performance with Reduced Resources
This paper introduces SAE-Rad, a novel approach using sparse autoencoders (SAEs) to generate interpretable radiology reports by decomposing image features into human-understandable concepts, achieving competitive performance with fewer resources compared to traditional VLMs.
A Radiologist-Minded Framework for Radiology Report Generation Across Anatomical Regions
This paper introduces X-RGen, a novel framework for generating radiology reports across multiple anatomical regions, mimicking the reasoning process of human radiologists to improve accuracy and clinical relevance.
Improving Radiology Report Generation and Evaluation Using Layman's Terms
This research proposes a novel framework called Layman's RRG, which leverages layman's terms to improve the generation and evaluation of radiology reports, addressing the limitations of traditional word-overlap metrics and the highly technical nature of medical language.
専門用語を使わないレントゲンレポート生成： より正確な評価と、より人間らしい解釈に向けて
従来のレントゲンレポート生成における評価指標は、専門用語の多用により、患者の理解を妨げ、モデルの学習を歪ませる可能性がある。本稿では、専門用語を使わない、わかりやすい言葉で記述されたレポート生成の枠組みを提案し、より正確な評価と、より人間らしい解釈の実現を目指す。