文献速递心血管胸部CT

基于心外膜脂肪组织CT影像组学分析风险因素对心房颤动类型的影响

Impact of risk factors on atrial fibrillation types via epicardial adipose tissue computed tomography-based radiomics analysis.
PubMed 文献 · Clin Radiol · 2024-11-21

本研究纳入2018年6月至2020年12月接受射频消融的房颤患者,经倾向性评分匹配后共794例(阵发性397例、持续性397例),采用nnU-Net分割心外膜脂肪组织并提取影像组学特征。结果显示持续性房颤的心外膜脂肪组织CT灰度值更高、纹理更紊乱,且部分影像组学特征与高血压、高脂血症、性别、年龄及糖尿病等风险因素相关。

要点

  • 持续性房颤的心外膜脂肪组织CT灰度值高于阵发性房颤,纹理更紊乱。
  • 15.11%的影像组学特征与房颤类型显著相关(P<.000005)。
  • 141个特征同时与风险因素和房颤类型显著相关,聚为三个独立簇。

核心结论

  • 心外膜脂肪组织影像组学特征在阵发性与持续性房颤间存在差异,持续性房颤表现为更高的CT灰度值和更紊乱的纹理。

研究设计与样本

  • 回顾性纳入2018年6月至2020年12月接受射频消融的房颤患者,经倾向性评分匹配后共794例(阵发性397例、持续性397例),采用nnU-Net分割心外膜脂肪组织并提取影像组学特征。

关键影像表现/征象

  • 持续性房颤的心外膜脂肪组织CT灰度值高于阵发性房颤。
  • 持续性房颤的心外膜脂肪组织纹理较阵发性房颤更紊乱。

诊断标准与关键数值

  • 单变量分析中15.11%的影像组学特征与房颤类型显著相关(P<.000005)。
  • 39.29%的特征与至少一个风险因素显著相关,其中141个特征同时与风险因素和房颤类型显著相关。
  • 141个特征经层次聚类识别出三个独立簇,最大簇以高血压和高脂血症因素为主,其对心外膜脂肪组织的影响高度重叠;其次是性别、年龄和糖尿病簇,共覆盖141个特征中的59个。

临床应用与阅片要点

  • 心外膜脂肪组织CT影像组学特征(灰度值、纹理)可能有助于区分阵发性与持续性房颤。
  • 高血压、高脂血症、性别、年龄及糖尿病等风险因素与心外膜脂肪组织影像组学特征相关,分析时需考虑其影响。

局限性

  • 原文未报告局限性。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

英文原文(摘要)· 点击展开

AIM: To analyze epicardial adipose tissue (EAT) radiomics and assess the impact of risk factors on different types of atrial fibrillation (AF).

MATERIALS AND METHODS: We included AF patients undergoing radiofrequency ablation from June 2018 to December 2020. Propensity score matching (PSM) was employed to limit the differences in risk factors between patients with different types of AF. The nnU-Net model was utilized to segment EAT. After extracting radiomic features, a generalized linear model was used to examine the epicardial adipose tissue (EAT) radiomic features associated with risk factors, and logistic regression analysis was conducted to identify the features differentiating the two AF groups. Hierarchical clustering was utilized to identify clusters among significant features.

RESULTS: After PSM, 794 patients (median age 67, 547 males) were selected, with 397 having paroxysmal and 397 persistent AF. In univariate analysis, 15.11% of the features were significantly associated with the type of AF (P<.000005). 39.29% of all features were significantly correlated with at least one risk factor, with 141 features being significantly related to both risk factors and the type of AF. Three independent clusters were successfully identified among the 141 features. The biggest cluster is dominated by hypertension and hyperlipidemia factors, exhibiting a high overlap in their effects on EAT. Following closely in influence are gender, age, and diabetes clusters, collectively covering 59 out of the 141 features.

CONCLUSION: Our study identifies distinct differences in EAT characteristics between paroxysmal and persistent AF, with persistent cases showing higher EAT gray values and a more disordered texture.

来源:Clin Radiol,查看原文。版权归原作者所有。

引用

  1. Hu Y, Zhou G, Jiang S, et al. Impact of risk factors on atrial fibrillation types via epicardial adipose tissue computed tomography-based radiomics analysis. Clin Radiol. 2024. PMID: 39689621. doi:10.1016/j.crad.2024.106753

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。