Department of Diagnostic and Interventional Radiology, University Hospital of Tübingen, Hoppe-Seyler-Str 3, Tübingen 72076, Germany.
Department of Radiology, NYU Grossman School of Medicine, New York, NY.
AJR Am J Roentgenol. 2022 Jun;218(6):1021-1029. doi: 10.2214/AJR.21.26922. Epub 2022 Jan 12.
Diagnosing liver lesions is challenging. CT is used for primary diagnosis, but its contrast resolution is limited. Investigating methods to improve detection of liver lesions is important. The purpose of this study was to evaluate the effect of frequency-selective nonlinear blending on the detectability of liver lesions on CT. A retrospective search yielded 109 patients with 356 malignant and benign liver lesions (191 principally diagnosed, 165 incidental findings) who underwent contrast-enhanced CT (CECT) in the portal venous phase and liver MRI between January 2012 and December 2017. Nonlinear blending was applied to CECT examinations, and three blinded readers independently rated the quality (5-point Likert scale) of randomly presented images. Focal lesions ( = 356) were evaluated for lesion identification and categorization to assess sensitivity. For 191 lesions (primary diagnosis), two readers evaluated CECT and nonlinear blending CT to compare lesion size and the accuracy of subjective measurements. A fourth reader performed ROI measurements for calculation of contrast-to-noise ratio (CNR), and a fifth reader reviewed MRI as the standard of reference. Statistics included interobserver agreement, quantitative comparisons of CNR, lesion size, and subjective image analyses of image quality and sensitivity for detecting liver lesions. Three readers rated the image quality of nonlinear blending CT (rating, 4; 10th-90th percentiles, 4-5) higher than that of CECT (rating, 2; 10th-90th percentiles, 1-3) ( < .001). CECT had good interreader agreement (interclass correlation coefficient [ICC], 0.81; 95% CI, 0.76-0.85), as did nonlinear blending CT (ICC, 0.75; 95% CI, 0.69-0.79). The median CNR of liver lesions increased with nonlinear blending (CECT, 4.18 [10th-90th percentiles, 1.67-9.06]; nonlinear blending CT, 12.49 [10th-90th percentiles, 6.18-23.39]; < .001). Bland-Altman analysis of lesion size showed a reduction in underestimation from 2.5 (SD, 9.2) mm (95% CI, 1.2-3.9 mm) with CECT to 0.1 (SD, 3.9) mm (95% CI, -0.68 to 0.46 mm) for nonlinear blending CT (concordance correlation coefficient, 0.99). Sensitivity for detecting liver lesions increased to 86% for nonlinear blending CT. The sensitivity of CECT was 76%. Frequency-selective nonlinear blending in CECT increases image quality and CNR, increases the precision of size measurement, and increases sensitivity for detecting liver lesions. Use of nonlinear blending CT improves liver lesion detection and increases the accuracy of lesion size measurement, which is important when local ablation or liver transplant is being considered.
诊断肝病变具有挑战性。CT 用于初步诊断，但对比度分辨率有限。研究提高肝病变检测能力的方法很重要。本研究旨在评估频率选择非线性混合对 CT 肝病变检测能力的影响。通过回顾性搜索，我们获得了 2012 年 1 月至 2017 年 12 月期间接受对比增强 CT（CECT）门静脉期和肝脏 MRI 检查的 109 例 356 例恶性和良性肝病变（191 例主要诊断，165 例偶然发现）患者的资料。对 CECT 检查应用非线性混合，三位盲法读者独立对随机呈现的图像质量（5 分 Likert 量表）进行评分。对焦点病变（ = 356）进行评价，以评估其对病变识别和分类的敏感性。对于 191 个病变（主要诊断），两位读者评估了 CECT 和非线性混合 CT，以比较病变大小和主观测量的准确性。第四位读者对感兴趣区域（ROI）进行测量，以计算对比噪声比（CNR），第五位读者将 MRI 作为参考标准进行回顾。统计学分析包括观察者间的一致性、CNR、病变大小和对检测肝病变的图像质量和敏感性的主观图像分析的定量比较。三位读者评价非线性混合 CT 的图像质量（评分，4；10 至 90 百分位，4-5）高于 CECT（评分，2；10 至 90 百分位，1-3）（ <.001）。CECT 具有良好的观察者间一致性（组内相关系数[ICC]，0.81；95%置信区间，0.76-0.85），非线性混合 CT 也具有良好的观察者间一致性（ICC，0.75；95%置信区间，0.69-0.79）。肝病变的中位 CNR 随非线性混合而增加（CECT，4.18 [10 至 90 百分位，1.67-9.06]；非线性混合 CT，12.49 [10 至 90 百分位，6.18-23.39]； <.001）。病变大小的 Bland-Altman 分析显示，CECT 从 2.5（标准差，9.2）mm（95%置信区间，1.2-3.9 mm）的低估减少到非线性混合 CT 的 0.1（标准差，3.9）mm（95%置信区间，-0.68 至 0.46 mm）（一致性相关系数，0.99）。非线性混合 CT 检测肝病变的敏感性提高至 86%。CECT 的敏感性为 76%。CECT 中的频率选择非线性混合可提高图像质量和 CNR，提高病变大小测量的精度，并提高肝病变的检测敏感性。非线性混合 CT 的使用可提高肝病变的检测能力，并提高病变大小测量的准确性，当考虑局部消融或肝移植时，这一点很重要。
