文献速递神经外周血管MR

病变大小对血管壁MRI定量测量与影像组学特征可重复性的影响

Impact of lesion size on reproducibility of quantitative measurement and radiomic features in vessel wall MRI.
PubMed 文献 · Eur Radiol · 2022-11-17

该单中心回顾性研究纳入251例颅内动脉病变患者,评估血管壁MRI定量测量与影像组学特征的观察者内可重复性,并按病变大小分层分析。结果显示,厚度测量与影像组学特征的重复性随病变厚度和体积增大而提高;基于强度的统计特征在最小、最薄病变中重复性最佳。

要点

  • 病变厚度≥2.5 mm或体积≥50 mm³时影像组学特征重复性更高(中位CCC 0.97 vs 0.89/0.88,P<.001)。
  • 基于强度的统计特征在最小、最薄病变中仍保持最高可重复性。
  • 定量测量与组学特征的可重复性随病变尺寸增大呈序列性升高。

局限性

  • 单中心回顾性设计可能限制结果的外推性。
  • 仅评估观察者内可重复性,未涉及观察者间可重复性。
  • 仅使用三维涡轮自旋回波T1加权成像序列,未涵盖其他血管壁MRI序列。

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

OBJECTIVES: To investigate reproducibility of quantitative measurement and radiomic features in vessel wall MRI (VW-MRI), evaluate the impact of lesion size, and identify reproducible radiomic features.

METHODS: This retrospective, single-center study included 251 patients (mean age, 53 ± 12 years; 128 women) with atherosclerosis, dissection, aneurysm, moyamoya disease, and vasculitis of the intracranial arteries who underwent three-dimensional turbo spin echo T1-weighted image. Lesion thickness, volume, and signal intensity were measured, and 157 radiomic features were extracted. Intra-observer reproducibility of quantitative measurement and radiomic features was evaluated by calculating the concordance correlation coefficient (CCC) and proportion of radiomic features above the predefined CCC. The reproducibility of quantitative measurement and radiomic features according to lesion size (binary comparison and stratification into 5 and 18 groups) was evaluated.

RESULTS: There was an overall serial increase in CCC for thickness measurement when stratified by lesion thickness and volume. There was an overall serial increase in the median CCC for radiomic features and proportion of radiomic features with CCC > 0.85 when stratified by lesion thickness and volume. Reproducibility of radiomic features was higher in the lesions with thickness ≥ 2.5 mm (median CCC, 0.97 vs. 0.89, p < .001; proportion with CCC > 0.85, 88.5% vs. 59.6%, p < .001) and volume ≥ 50 mm3 (median CCC, 0.97 vs. 0.88, p < .001; proportion with CCC > 0.85, 90.4% vs. 59.0%, p < .001). Intensity-based statistical features remained most reproducible in the thinnest and smallest lesions.

CONCLUSIONS: Intra-observer reproducibility of thickness measurement and radiomic features was affected by lesion size in VW-MRI although intensity-based statistical features remained most reproducible.

KEY POINTS: • There was an overall serial increase in CCC for thickness measurement when stratified by lesion size. • There was an overall serial increase in the median CCC for radiomic features and proportion of radiomic features with CCC > 0.85 when stratified by lesion size. • Intensity-based statistical features remained most reproducible in the thinnest and smallest lesions.

来源:Eur Radiol,查看原文。版权归原作者所有。

引用

  1. Kim, Jung, Park, et al. Impact of lesion size on reproducibility of quantitative measurement and radiomic features in vessel wall MRI. Eur Radiol. 2022. PMID: 36394600. doi:10.1007/s00330-022-09207-2

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。