From the Department of Biophysics and Nuclear Medicine, Bicêtre University Hospital, Assistance Publique-Hôpitaux de Paris, 78 rue du Général Leclerc, 94275 Le Kremlin-Bicêtre, France (F.L.B., T.H., C.M., V.C., V.R., G.G., M.C., P.C.R., E.D.); IR4M-UMR 8081, Université Paris Saclay, Université Paris Sud, CNRS, Orsay, France (F.L.B., E.D.); Service Hospitalier Frédéric Joliot, CEA, Université Paris-Sud, Orsay, France (E.B., V.A., V.L.); Department of Nuclear Medicine, Hôpital Marie Lannelongue, Le Plessis Robinson, France (L.M.); Université Paris-Sud, Faculté de Médecine, Université Paris-Saclay, Le Kremlin-Bicêtre, France (F.P., A.S., S.B., D.M., M.H.); Service de Pneumologie, Hôpital Bicêtre, AP-HP, Le Kremlin-Bicêtre, France (F.P., A.S., S.B., D.M., M.H.); and INSERM UMR_S 999, Hôpital Marie Lannelongue, Le Plessis Robinson, France (F.P., A.S., S.B., D.M., M.H.).
Radiology. 2018 Jul;288(1):277-284. doi: 10.1148/radiol.2018171756. Epub 2018 Apr 3.
Purpose To assess the performance of the ITK-SNAP software for fluorodeoxyglucose (FDG) positron emission tomography (PET) segmentation of complex-shaped lung tumors compared with an optimized, expert-based manual reference standard. Materials and Methods Seventy-six FDG PET images of thoracic lesions were retrospectively segmented by using ITK-SNAP software. Each tumor was manually segmented by six raters to generate an optimized reference standard by using the simultaneous truth and performance level estimate algorithm. Four raters segmented 76 FDG PET images of lung tumors twice by using ITK-SNAP active contour algorithm. Accuracy of ITK-SNAP procedure was assessed by using Dice coefficient and Hausdorff metric. Interrater and intrarater reliability were estimated by using intraclass correlation coefficients of output volumes. Finally, the ITK-SNAP procedure was compared with currently recommended PET tumor delineation methods on the basis of thresholding at 41% volume of interest (VOI; VOI) and 50% VOI (VOI) of the tumor's maximal metabolism intensity. Results Accuracy estimates for the ITK-SNAP procedure indicated a Dice coefficient of 0.83 (95% confidence interval: 0.77, 0.89) and a Hausdorff distance of 12.6 mm (95% confidence interval: 9.82, 15.32). Interrater reliability was an intraclass correlation coefficient of 0.94 (95% confidence interval: 0.91, 0.96). The intrarater reliabilities were intraclass correlation coefficients above 0.97. Finally, VOI and VOI accuracy metrics were as follows: Dice coefficient, 0.48 (95% confidence interval: 0.44, 0.51) and 0.34 (95% confidence interval: 0.30, 0.38), respectively, and Hausdorff distance, 25.6 mm (95% confidence interval: 21.7, 31.4) and 31.3 mm (95% confidence interval: 26.8, 38.4), respectively. Conclusion ITK-SNAP is accurate and reliable for active-contour-based segmentation of heterogeneous thoracic PET tumors. ITK-SNAP surpassed the recommended PET methods compared with ground truth manual segmentation.
目的 评估 ITK-SNAP 软件在复杂形状肺部肿瘤的氟脱氧葡萄糖（FDG）正电子发射断层扫描（PET）分割中的性能，与优化的、基于专家的手动参考标准相比。
材料和方法 回顾性地对 76 例胸部病变的 FDG PET 图像使用 ITK-SNAP 软件进行分割。每个肿瘤均由 6 名评估者手动分割，以使用同时真实性和性能水平估计算法生成优化的参考标准。4 名评估者使用 ITK-SNAP 主动轮廓算法对 76 例肺部肿瘤的 FDG PET 图像进行了两次分割。使用 Dice 系数和 Hausdorff 度量评估 ITK-SNAP 程序的准确性。通过输出体积的组内相关系数评估组内和组间的可靠性。最后，根据感兴趣区域（VOI）的 41%（VOI）和肿瘤最大代谢强度的 50%（VOI）阈值，基于 ITK-SNAP 程序与目前推荐的 PET 肿瘤勾画方法进行比较。
结果 ITK-SNAP 程序的准确性估计值为 Dice 系数为 0.83（95%置信区间：0.77，0.89），Hausdorff 距离为 12.6mm（95%置信区间：9.82，15.32）。组间可靠性为组内相关系数 0.94（95%置信区间：0.91，0.96）。组内可靠性的组内相关系数均高于 0.97。最后，VOI 和 VOI 的准确性指标如下：Dice 系数分别为 0.48（95%置信区间：0.44，0.51）和 0.34（95%置信区间：0.30，0.38），Hausdorff 距离分别为 25.6mm（95%置信区间：21.7，31.4）和 31.3mm（95%置信区间：26.8，38.4）。
结论 ITK-SNAP 软件在基于主动轮廓的异质性胸部 PET 肿瘤分割中是准确和可靠的。与基于地面实况的手动分割相比，ITK-SNAP 优于推荐的 PET 方法。
