孤立性肺结节^(18)FDG PET/CT诊断的数学模型建立

Establishment of a Mathematical Model for Diagnosis of the SPN with ^(18)FDG PET/CT

目的:利用二元Logistic回归法建立18FDG PET/CT诊断孤立性肺结节(SPN)良、恶性的数学模型。方法:接受PET/CT检查的连续182例SPN患者(良性67例、恶性115例)纳入研究,选择患者年龄、性别以及病灶部位、最大径、密度、瘤肺界面、分叶、毛刺、血管集束征、胸膜牵拉征和FDG摄取程度作为诊断的影响因子进行单因素和多因素分析,应用二元Logistic回归法建立SPN的PET/CT定性诊断数学模型。结果:SPN定性诊断的Logistic数学模型为p=ex/(1+ex),x=-4.146+0.041×年龄+2.226×密度-1.053×瘤肺界面+1.211×分叶+2.579×血管集束征+1.954×胸膜牵拉征+0.286×SUVmax。数学模型区分SPN良、恶性的ROC曲线下面积(AUC)为0.889±0.025,显著高于单纯SUVmax(AUC=0.729±0.038,P<0.05)。结论:利用二元Logistic回归建立的18FDG PET/CT区分SPN良、恶性的数学模型有很高的诊断准确率。

Purpose： To establish a mathematical model for diagnosis of the solitary pulmonary nodules （SPN） with binary Logistic regression analysis. Methods： One hundred and eighty-two patients with SPN （115 malignant, 67 benign） were collected in the study. Clinical data included 12 items （age, gender, maximum diameter, site, density, interface of tumor and parenchyma, lobulation, speculation, pleural retraction sign, vascular convergence and FDG uptake） were analyzed by univariate and multivariate statistical method. The mathematical model was obtained from binary Logistic regression. Results： The mathematical model established by binary Logistic regression was： p=eX/（ 1＋ eX）, x= -4.146＋0.041~ age＋2.226 x density -1.053 x （interface of tumor and parenchyma） ＋1.211 ~lobulation ＋2.579 ~ （vascular convergence） ＋1.954~ （pleural retraction sign） ＋0.286 ~ SUVmax. The AUC of mathematical model was 0.889_＋ 0.025 which was better than the AUC of SUVmax（AUC=0.729_＋0.038, p〈0.05）. Conclusion： The mathematical model established by binary Logistic regression has high diagnostic value for diagnosis of SPN and will be used in clinical practice well.