Hernandez Andrew M, Chen Anthony F, Sen Fatma, Mitchell Ana S, McKenney Sarah E, Nardo Lorenzo, Abbey Craig K, Madani Mohammad H
Department of Radiology, University of California Davis, Sacramento, California, United States.
Department of Psychological and Brain Sciences, University of California, Santa Barbara, California, United States.
J Clin Imaging Sci. 2025 Jul 8;15:25. doi: 10.25259/JCIS_17_2025. eCollection 2025.
The objective of the study was to evaluate the efficacy of ultra-high-resolution computed tomography (UHRCT) in comparison to conventional resolution computed tomography (CT) for the characterization of lung nodules.
104 non-contrast chest UHRCT scans (mean age of 66 years, 57 females) with pulmonary nodules were retrospectively collected (February-November 2022), and corresponding normal-resolution (NR) reconstructions were synthesized using a validated algorithm. Five blinded radiologists scored the following for each localized nodule in the ultra-high-resolution (UHR) and NR datasets: Margin clarity (5-point Likert scale), image quality "IQ" (3-point), density confidence (0-100%), and size (long/short axes). Image noise (voxel standard deviation) was calculated within the trachea. Differences between UHR and NR were tested using the Wilcoxon signed-rank test. Intrareader agreement was quantified with intraclass correlation coefficient (ICC), and ordinal association between margin clarity and IQ was quantified with Kendall's coefficient.
Margin clarity, IQ, and density confidence were significantly higher for UHR ( < 0.001). No significant differences between UHR and NR were observed in the variability (standard deviation across readers) for measuring long and short axes ( > 0.100). Intrareader agreement for UHR and NR was poor for margin clarity, IQ, and density confidences (ICC < 0.250) but moderate for short axes (ICC = 0.731) and good for long axes (ICC = 0.807). Ordinal association between margin clarity and IQ was moderate for UHR ( = 0.566) and good for IQ ( = 0.637). Image noise was significantly higher ( < 0.001) for UHR compared to NR.
UHRCT offers significant improvements in the visualization of lung nodules compared to conventional resolution CT, albeit with an increase in image noise.
本研究的目的是评估超高分辨率计算机断层扫描（UHRCT）与传统分辨率计算机断层扫描（CT）相比在肺结节特征描述方面的有效性。
回顾性收集了104例有肺结节的非增强胸部UHRCT扫描（平均年龄66岁，57例女性）（2022年2月至11月），并使用经过验证的算法合成了相应的正常分辨率（NR）重建图像。五名不知情的放射科医生对超高分辨率（UHR）和NR数据集中的每个局部结节进行以下评分：边缘清晰度（5级李克特量表）、图像质量“IQ”（3级）、密度置信度（0 - 100%）和大小（长/短轴）。在气管内计算图像噪声（体素标准差）。使用Wilcoxon符号秩检验来检验UHR和NR之间的差异。用组内相关系数（ICC）对阅片者内一致性进行量化，用肯德尔系数对边缘清晰度和IQ之间的有序关联进行量化。
UHR的边缘清晰度、IQ和密度置信度显著更高（<0.001）。在测量长轴和短轴的变异性（阅片者间标准差）方面，未观察到UHR和NR之间存在显著差异（>0.100）。UHR和NR在边缘清晰度、IQ和密度置信度方面的阅片者内一致性较差（ICC<0.250），但在短轴方面为中等（ICC = 0.731），在长轴方面为良好（ICC = 0.807）。UHR的边缘清晰度和IQ之间的有序关联为中等（=0.566），IQ方面为良好（=0.637）。与NR相比，UHR的图像噪声显著更高（<0.001）。
与传统分辨率CT相比，UHRCT在肺结节可视化方面有显著改善，尽管图像噪声有所增加。
