Effect Of Different Reconstruction Parameters On Volumetric Measurement Of Pulmonary Nodules In64-slice Spiral CT

Objective To evaluate the effect of different reconstruction algorithms and slice thicknesses on volumetric measurement of pulmonary nodules in64-slice spiral CT.Materials and Methods Sixty-three patients were collected who underwent64-slice spiral CT scan in Tianjin Medical University General Hospital due to the follow-up of indeterminate pulmonary nodules from December2010to September2011. The criteria for pulmonary nodules included were solid nodules with a diameter less than15mm in diameter and no connection with the pleural or vessels, no calcification or cavity.In total,30patients (male16, female14) with62nodules were enrolled, aged from35to78years, and the mean age is58.6±8.6years. All CT data were reconstructed with six reconstruction algorithms(soft, standard, chest, lung, bone, edge) and three slice thicknesses(0.625mm、1.25mm、2.5mm),resulting18groups of data. All18groups of data were transported to GE AW4.4workstation. The volumetric measurements of the nodules were performed with advanced lung analysis (ALA) software by one observer. The volume and diameter of the nodules were recorded. A repeated measurement by the same observer was performed one week later, using the data reconstructed with standard algorithms and3different slice thicknesses. The effect of reconstruction algorithms, slice thickness and nodule diameter on volumetric measurement were assessed by multivariate analysis of variance for repeated measures, correlation analysis and Bland-Altman method. The intra-observer variability when using standard algorithm was evaluated as well.Results Reconstruction algorithms (F=10.4, P<0.001) and slice thicknesses (F=5.5, P=0.006) had significant effect on measured volume of pulmonary nodules. Standard algorithm had significant difference with chest, lung bone and edge algorithm, and so was soft algorithm. There was significant difference between chest algorithm and bone algorithm and between2.5mm slice thickness and each of the other two slice thicknesses. The coefficients of variation of18measurements was inversely related with nodule diameter, with a negative coefficient of correlation as-0.822(P<0.001). The volume measured at2.5mm slice thickness had a poor agreement with that at1.25mm and0.625mm, respectively. The best agreement between1.25mm and 0.625mm was achieved when bone algorithm was used. No matter what the reconstruction settings were, the agreement of nodules≥5mm was better than that of nodules<5mm. The best intra-observer repeatability was achieved when0.625mm slice thickness was used for nodules≥5mm. The intra-observer repeatability was the worst when we used2.5mm reconstruction slice thickness for nodules<5mm.Conclusion Reconstruction algorithms and slice thicknesses have significant impact on lung nodule volume measurement, especially for small nodules. In the follow-up of indeterminate pulmonary nodules, reconstruction algorithms and slice thicknesses in serial CT scans should be consistent, which is more important for small nodules. Especially if you used standard or soft algorithm in the first CT scan, you should not use any kind of high resolution algorithms in the follow-ups, or it may lead to misjudgment of the percentage volume change of the nodule. For nodules<5mm,2.5mm reconstruction slice thickness is not proper, and we should use1.25mm or smaller reconstruction slice thickness instead.