数值逼近法确定PET图像肿瘤靶区-动物模型的评价

General Target Volume Delineation on PET images by a Numerical Approximation Method – Evaluation by an Animal Study

介绍了采用数值逼近法确定PET图像的肿瘤靶区-动物模型。运用VX2荷瘤兔肿瘤模型验证了数值逼近法设计的SUV_Shape程序测量肿瘤大体体积的准确性。肿瘤体积采用椭球体体积计算公式、CT阈值法、SUV_Shape计算,配对t检验及相关性分析结果表明:VX2肿瘤12个肿瘤大体体积3种方法测量结果之间无统计学差异(t≤1.14,P≥0.28);且高度相关(r≥0.894,P<0.01);运用SUV_Shape程序,根据PET图像测量肿瘤大体体积具有较高的准确性。

An animal study was introduced that delineated general tumor volumes on PET images by a numerical approximation method. A scheme named SUV_Shape was developed by the numerical approximation method and evaluated by VX.2 rabbit models. The VX2 tumor node or masses were measured by the ellipsoid volume formula, high resolution CT scan, and SUV_Shape. The paired -t tests and Pearson＇s correlation coefficient were done. 12 VX2 tumors were analyzed. Their GTV were not significant different （ t ≤1.14, P I〉 0.28 ） , and highly correlated （ r 〉 0.89, P 〈 0.01 ） among three measurements. The accuracy of GTV measurement on PET images by SUV_Shape was high.