Data from: Reproducibility of volumetric computed tomography of stable small pulmonary nodules with implications on estimated growth rate and optimal scan interval

Purpose: To use clinically measured reproducibility of volumetric CT (vCT) of lung nodules to estimate error in nodule growth rate in order to determine optimal scan interval for patient follow-up. Methods: We performed quantitative vCT on 89 stable non-calcified nodules and 49 calcified nodules measuring 3–13 mm diameter in 71 patients who underwent 3–9 repeat vCT studies for clinical evaluation of pulmonary nodules. Calculated volume standard deviation as a function of mean nodule volume was used to compute error in estimated growth rate. This error was then used to determine the optimal patient follow-up scan interval while fixing the false positive rate at 5%. Results: Linear regression of nodule volume standard deviation versus the mean nodule volume for stable non-calcified nodules yielded a slope of 0.057±0.002 (r2 = 0.79, p<0.001). For calcified stable nodules, the regression slope was 0.052±0.005 (r2 = 0.65, p = 0.03). Using this with the error propagation formula, the optimal patient follow-up scan interval was calculated to be 81 days, independent of initial nodule volume. Conclusions: Reproducibility of vCT is excellent, and the standard error is proportional to the mean calculated nodule volume for the range of nodules examined. This relationship constrains statistical certainty of vCT calculated doubling times and results in an optimal scan interval that is independent of the initial nodule volume.目的：采用临床测量得到的肺结节容积CT（volumetric CT，vCT）可重复性，评估结节生长速率估计值的误差，进而确定患者随访的最佳扫描间隔。方法：本研究对71例因肺结节临床评估而行3~9次重复容积CT扫描的患者的89个稳定非钙化结节与49个钙化结节（直径3~13 mm）开展定量容积CT分析。以结节平均体积为自变量计算体积标准差，以此估算生长速率估计值的误差；在固定假阳性率为5%的前提下，利用该误差确定患者随访的最佳扫描间隔。结果：针对稳定非钙化结节，结节体积标准差与平均结节体积的线性回归分析得到斜率为0.057±0.002（决定系数r²=0.79，p<0.001）；对于稳定钙化结节，回归斜率为0.052±0.005（r²=0.65，p=0.03）。结合误差传播公式计算后，得到患者随访的最佳扫描间隔为81天，且该间隔与结节初始体积无关。结论：在本研究纳入的肺结节范围内，容积CT的可重复性极佳，且标准误差与计算得到的平均结节体积呈正相关。该关系限制了容积CT计算所得结节倍增时间的统计置信度，并最终得到与结节初始体积无关的最佳扫描间隔。