重建算法和层厚对肺结节容积定量的影响

Impacts of reconstruction kernel and section thickness on volumetric measurement

目的用肺结节模型评价多排探测器CT(MDCT)的重建算法、层厚对不同大小、密度肺结节容积定量准确性的影响,并对肺结节容积定量的重复性进行检验。方法肺结节模型直径25cm,内含4种直径(2.5、5.0、10.0、20.0mm)、5种密度(-100、-60、0、60、100HU)结节共16个。利用64排MDCT对模型进行两次扫描,采用8种重建算法(软组织、标准、胸、肺、精细、骨、骨+、边缘)和五种层厚(0.625、1.25、2.5、3.75、5.0mm),获取评估结节定量准确性的数据和用于重复性检验的数据。在ADW4.2工作站上用ALA(advancedlung analysis)软件对模拟肺结节进行容积测量。用CT软件测量获得的容积相对于参考标准容积的绝对误差百分率(APE)表示容积定量的准确性。统计学方法采用单因素及多因素方差分析、多元线性回归分析及配对t检验。结果①在0.625mm及1.25mm层厚条件下,8种重建算法APE之间的差异有统计学意义(P<0.05),其中肺算法误差最大,且与其他7种算法之间的差异均有统计学意义(P<0.05);其他7种算法间差异无统计学意义(P>0.05),其中胸算法误差最小,骨算法次之。②重建层厚、结节直径及结节密度对结节容积APE值的影响均有统计学意义(P<0.01)。对于直径为2.5mm的结节,仅能用0.625mm层厚测量且误差较大;对于直径≥5mm的结节,1.25mm层厚的测量误差最小,与0.625mm层厚的结果相比差异无统计学意义(P>0.05)。重建层厚(P<0.001)、结节直径(P<0.001)与APE值存在线性回归关系。③当采用0.625mm和1.25mm两种层厚重建时,两次扫描容积定量值之间差异无统计学意义(P>0.05)。结论除肺算法外的其余7种重建算法均可用于肺结节容积定量;对于直径<5mm的结节,应采用0.625mm或更薄重建层厚,对于直径≥5mm的结节,以采用1.25mm重建层厚为宜;结节直径和(或)密度的变化对容积APE值也具有显著影响;MDCT肺结节容积定量的重复性可靠。

Objective To evaluate the impacts of reconstruction kernel and section thickness on volumetric measurement of pulmonary nodule with different nodule size and density in a phantom using multi detector CT （MDCT）, and to test the reproducibility of volume measurements of pulmonary nodules. Methods The diameter of phantom was 25 cm, a total of 16 contained spherical synthetic nodules with 4 diameter categories （2.5 ram, 5.0 mm, 10.0 mm, 20.0 mm）, and 5 attenuation categories （-100 HU, -60 HU, 0 HU, 60 HU, 100 HU） were studied. The phantom was scanned twice at intervals of two weeks with 64-detector CT. Eight reconstruction kernels （soft tissue, standard, chest, lung, detail, bone, bone＋, edge） and 5 section thicknesses （0. 625 mm, 1.25 mm, 2.5 mm, 3.75 mm, 5.0 mm） were used to obtain data for assessing accuracy and the reproducibility of volumetric measurement. The image data were transferred to ADW 4.2 workstation, and the volumes of lung nodules were measured with ALA （advanced lung analysis） software package. Absolute percentage error （APE） was calculated to assess the accuracy of volumetric measurement. Statistical analysis was performed using one-way ANOVA, multivariate analysis of variance, multiple linear regression analysis and paired-samples t test. Results （1）There was statistically significant difference in the APE across the 8 reconstruction kernels with the 0. 625 mm and 1.25 mm section thicknesses （P〈0. 05）. Lung kernel yielded the maximal error, and there were statistically significant differences in comparison with the other seven kernels （P〈0.05）. However, there was no significant difference between the rest seven kernels （P〉0. 05）, with the chest kernel yielding the minimal error,and the bone kernel followed. （2）Reconstruction section thickness, nodule size and nodule attenuation signifieantly affected the APE （P〈0.01）. For 2.5 mm nodules, the volume could be only measured on 0. 625 mm section thickness, and the error was greater. For nodules 5 mm or larger, measurements obtained by using the 1.25 mm section thickness yielded the minimal error, although no significant difference was found compared with 0. 625 mm sections. Both the reconstruction section thickness and nodule diameter had linear correlation with APE. （3）There was no significant variation in the volumes obtainded by chest kernel with the section thicknesses of 0. 625 mm and 1.25 mm between the twice scans （P〉0.05）. Conclusion All kernels except the lung kernel can be used for volumetric measurement. For nodules less than 5 mm, the section thickness should be 0. 625 mm or less, while for nodules 5 mm or larger, a reconstruction section thickness of 1.25 mm is recommended. The varying of nodule diameter and/or density also significantly affects volumetric APE. The reproducibility of MDCT-based volumetric measurements of pulmonary nodules is reliable.