不同影像引导方式下经皮肺穿刺活检肺出血 风险预测模型的研究现状与差异化分析

Research Status and Differential Analysis of Risk Prediction Models for Pulmonary Hemorrhage in Percutaneous Lung Biopsy under Different Imaging Guidance Methods

摘要: 经皮肺穿刺活检是肺部结节定性诊断的金标准，但肺出血风险在不同影像引导技术(CT、超声、CBCT、PET/CT等)下呈现显著异质性。通过回顾分析现有文献发现，CT引导下出血预测模型最为成熟，已从临床特征分析发展至影像组学与机器学习阶段；超声模型侧重血流动力学评估；PET/CT与MRI则通过优化靶点规划间接降低风险。相比之下，CBCT引导技术虽普及迅速，却缺乏专用量化风险预测模型，临床常直接套用CT模型，忽视了其特有操作流程与图像特征带来的风险差异。研究表明，单一通用模型难以满足精准医疗需求，未来需要构建CBCT引导下的特异性风险预测模型，并探索多模态数据融合的人工智能辅助决策系统。本文综述了不同影像引导技术下经皮肺穿刺活检肺出血风险预测模型的研究现状，分析其方法学特征及预测因子的异同，为构建适应其他新技术的精准预测模型提供了理论依据。

Abstract: Percutaneous lung biopsy is the gold standard for the definitive diagnosis of pulmonary nodules, but the risk of pulmonary hemorrhage varies significantly across different imaging guidance techniques (CT, ultrasound, CBCT, PET/CT, etc.). A review of the existing literature shows that bleeding prediction models under CT guidance are the most mature, evolving from analyses based on clinical features to radiomics and machine learning approaches; ultrasound models focus on hemodynamic assessment; PET/CT and MRI reduce risk indirectly by optimizing target planning. In contrast, although CBCT-guided techniques are rapidly becoming widespread, there is a lack of dedicated quantitative risk prediction models, and clinical practice often directly applies CT models, overlooking the risk differences brought by its unique procedural workflows and imaging characteristics. Studies indicate that a single universal model is insufficient for the needs of precision medicine. In the future, it will be necessary to develop specific risk prediction models under CBCT guidance and explore AI-assisted decision-making systems that integrate multimodal data. This article reviews the current research status of pulmonary hemorrhage risk prediction models in percutaneous lung biopsy under different imaging guidance techniques, analyzes their methodological characteristics and predictive factors, and provides a theoretical basis for developing precise prediction models adaptable to other emerging technologies.