Scholarlink

<p>This study aims at documenting the technical performances of the Skeletal Muscle Index at the L3 vertebra (L3SMI). In this retrospective multicenter study, we used TotalSegmentator freeware to segment muscles in axial CT scans at the L3 vertebra level and measure Skeletal Muscle Area (SMA). A dataset of 100 cancer patients (lung, colorectal, adrenal, pancreatic) was analyzed for accuracy, reproducibility, and linearity, using expert-segmented ground truth. A second dataset of 205 prone/supine colorectal cancer patients assessed repeatability in a test–retest setting. Outliers were reviewed to define exclusion criteria for L3SMI assessment. We computed within Coefficient of Variation (wCV), Bland–Altman analysis for variability and repeatability, and measured absolute/relative bias with confidence intervals from bootstrapped data. Bablok regression assessed measurement linearity. Intra and inter readers variability were, respectively, 5% and 7.6%. Relative bias was − 8.5% (95%CI: − 9.8, − 7.3), absolute bias was − 11.3cm<sup>2</sup> (95%CI: − 13.2, − 9.4) indicating the need for linear correction. The linear relationship between measurement and ground truth was found with a Pearson’s coefficient of 0.97. Repeatability gives wCV of 3.2% (95%CI: 2.8; 3.5), with decreasing wCV value of [4.0; 2.7] % for MSA ranges [66.3; 213.9] cm<sup>2</sup>. L3SMI evaluations were not applicable to ill positioned patients, truncated Field of View for steatosis patients in non-enhanced imaging. L3SMI and longitudinal changes of L3SMI can be reliably measured. Exclusion criteria have been listed.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Technical performance of L3 skeletal muscle area (SMA) measurement on CT for L3 skeletal muscle index (L3SMI) assessment

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月12日

查看全文
  • Hubert Beaumont,
  • Emilie Khayat,
  • Alexandre Thinnes,
  • Antoine Iannessi

摘要

This study aims at documenting the technical performances of the Skeletal Muscle Index at the L3 vertebra (L3SMI). In this retrospective multicenter study, we used TotalSegmentator freeware to segment muscles in axial CT scans at the L3 vertebra level and measure Skeletal Muscle Area (SMA). A dataset of 100 cancer patients (lung, colorectal, adrenal, pancreatic) was analyzed for accuracy, reproducibility, and linearity, using expert-segmented ground truth. A second dataset of 205 prone/supine colorectal cancer patients assessed repeatability in a test–retest setting. Outliers were reviewed to define exclusion criteria for L3SMI assessment. We computed within Coefficient of Variation (wCV), Bland–Altman analysis for variability and repeatability, and measured absolute/relative bias with confidence intervals from bootstrapped data. Bablok regression assessed measurement linearity. Intra and inter readers variability were, respectively, 5% and 7.6%. Relative bias was − 8.5% (95%CI: − 9.8, − 7.3), absolute bias was − 11.3cm2 (95%CI: − 13.2, − 9.4) indicating the need for linear correction. The linear relationship between measurement and ground truth was found with a Pearson’s coefficient of 0.97. Repeatability gives wCV of 3.2% (95%CI: 2.8; 3.5), with decreasing wCV value of [4.0; 2.7] % for MSA ranges [66.3; 213.9] cm2. L3SMI evaluations were not applicable to ill positioned patients, truncated Field of View for steatosis patients in non-enhanced imaging. L3SMI and longitudinal changes of L3SMI can be reliably measured. Exclusion criteria have been listed.