Ferraioli Giovanna, Roccarina Davide, Kumada Takashi, Ogawa Sadanobu, Yoshida Yuichi, Hirooka Masashi, Barr Richard G
Department of Clinical, Surgical, Diagnostic and Pediatric Sciences, University of Pavia, Campus della Salute, Pavia, Italy.
SOD Medicina Interna ed Epatologia, Azienda Ospedaliero-Universitaria Careggi, Firenze, Italy.
Eur Radiol. 2025 Sep 5. doi: 10.1007/s00330-025-11906-5.
Methods for measuring the ultrasound attenuation coefficient (AC) vary across different systems. Some have fixed regions of interest (ROI) while others have movable ROIs. Aims were to evaluate whether, using a system with a fixed ROI, correlation between AC and MRI proton density fat fraction (MRI-PDFF), and performance could be improved by (i) reducing fixed ROI length to 30 mm, changing starting point from the transducer, and (ii) using a movable ROI at different depths.
In this retrospective multicenter study, AC measurements were performed with the Arietta 850 system, and raw data were automatically stored. AC values were obtained using a standard commercial algorithm (ROI-setting1, 35-75 mm from transducer). Raw data were successively reprocessed externally using a fixed 45-75 mm ROI (ROI-setting2) and a movable 30 mm ROI positioned with the top at 20 mm (ROI-setting3) and 25 mm below the liver capsule (ROI-setting4). Spearman rho and area under the receiver operating characteristics curve (AUROC) were used to assess correlation with MRI-PDFF and performance, respectively, and the Delong test was used to compare AUROCs.
Seven hundred fifty participants (median age: 65 [52, 73] years; 384 males) were included. Correlation of ROI-setting1 with MRI-PDFF was 0.75 (0.72, 0.78), reaching 0.80 (0.77, 0.82) with ROI-setting4. Overall, ROI-setting4 showed significantly the best performance across steatosis grades. AUROCs for S > 0 were 0.90 (0.87, 0.92) for ROI-setting1 and 0.92 (0.90, 0.94) with ROI-setting4 (p < 0.001). This latter performed significantly better than all other settings in participants with obesity and skin-to-liver distance > 25 mm (p < 0.05).
A movable ROI improves both AC correlation with MRI-PDFF and performance. The highest improvement was with the ROI top 25 mm below the liver capsule.
Question It remains uncertain whether utilizing an ROI at a fixed position from the transducer is optimal for ultrasound AC measurements. Findings The highest AC performance and correlation with MRI-PDFF were observed with a movable ROI positioned with the top at 25 mm under the liver capsule. Clinical relevance A movable ROI with consistent sub-capsular placement improves both correlation between AC and MRI-PDFF, as well as its performance, whereas in individuals with a skin-to-liver capsule distance > 25 mm, a fixed ROI position significantly decreases AC performance.
不同系统测量超声衰减系数（AC）的方法各不相同。有些系统具有固定的感兴趣区域（ROI），而另一些则具有可移动的ROI。本研究旨在评估，对于使用固定ROI的系统，通过以下方式能否改善AC与MRI质子密度脂肪分数（MRI-PDFF）之间的相关性及性能：（i）将固定ROI长度缩短至30mm，改变换能器的起始点；（ii）在不同深度使用可移动ROI。
在这项回顾性多中心研究中，使用Arietta 850系统进行AC测量，并自动存储原始数据。使用标准商业算法（ROI设置1，距换能器35 - 75mm）获取AC值。原始数据随后在外部依次使用固定的45 - 75mm ROI（ROI设置2）和顶部位于肝包膜下方20mm（ROI设置3）及25mm（ROI设置4）的30mm可移动ROI进行重新处理。分别使用Spearman秩相关系数和受试者操作特征曲线下面积（AUROC）评估与MRI-PDFF的相关性及性能，并使用德龙检验比较AUROC。
纳入750名参与者（中位年龄：65[52, 73]岁；384名男性）。ROI设置1与MRI-PDFF的相关性为0.75（0.72, 0.78），ROI设置4时达到0.80（0.77, 0.82）。总体而言，ROI设置4在各脂肪变性分级中表现出显著最佳性能。S>0时，ROI设置1的AUROC为0.90（0.87, 0.92），ROI设置4为0.92（0.90, 0.94）（p<0.001）。在肥胖且皮肤至肝脏距离>25mm的参与者中，后者的表现明显优于所有其他设置（p<0.05）。
可移动ROI可改善AC与MRI-PDFF的相关性及性能。在肝包膜下方25mm处的ROI改善最为显著。
问题 对于超声AC测量，从换能器固定位置使用ROI是否最佳仍不确定。发现 观察到在肝包膜下方25mm处设置可移动ROI时，AC性能最高且与MRI-PDFF相关性最佳。临床意义 一致的包膜下放置可移动ROI可改善AC与MRI-PDFF之间的相关性及其性能，而在皮肤至肝包膜距离>25mm的个体中，固定ROI位置会显著降低AC性能。
