Comparison of rigid and deformable image registration for nasopharyngeal carcinoma radiotherapy planning with diagnostic position PET/CT

Comparison of rigid and deformable image registration for nasopharyngeal carcinoma radiotherapy planning with diagnostic position PET/CT
复制标题

DOI:
10.1007/s11604-019-00911-6
复制
发表时间:
2020-03-01
影响因子:
2.1
通讯作者:
Oya, Natsuo
Oya, Natsuo
中科院分区:
医学4区
文献类型:
--
作者:
Kai, Yudai;Arimura, Hidetaka;Oya, Natsuo

文献摘要

被引文献

相似文献

目的比较刚性图像配准(RIR)和变形图像配准(DED)在鼻咽癌(NPC)放射治疗计划中用于诊断位(DP)正电子发射断层扫描/计算机断层扫描(PET/CT)图像勾画大体肿瘤体积(GTV)的效果。材料与方法4名放射肿瘤学家分别使用RIR和基于B样条的重建方法在与DP-PET/CT图像配准的计划CT(pCT)图像上描绘GTV、GTV(RIR)和GTV(RIR)。所有放射肿瘤学家使用放射治疗位置(RP)-PET/CT图像独立描绘参考GTV。14例NPC患者分别采用早期和延迟扫描获得DP和RP-PET/CT图像。Dice的相似系数(DSC),平均距离协议,和体积协议与参考GTV进行了比较,考虑观察者间的变异性参考轮廓。结果GTV(RIR)和GTV(RIR)的DSC平均值分别为0.77和0.77,可用于GTV的勾画。GTV(RIR)与GTV(RIR)在各项评价指标上均无显著性差异(p > 0.05)。此外,颈部屈曲角度差异与GTV准确性之间的相关性没有统计学意义(p > 0.05)。结论RIR与B样条曲线拟合法相比,在不同颈前屈角度条件下对鼻咽癌GTV的勾画是一种可行的选择。
Purpose This observer study aimed to compare rigid image registration (RIR) with deformable image registration (DIR) for diagnostic position (DP) positron emission tomography/computed tomography (PET/CT) images in the delineation of gross tumor volumes (GTVs) in nasopharyngeal carcinoma (NPC) radiotherapy planning. Materials and methods Four radiation oncologists individually delineated the GTVs, GTV(RIR), and GTV(DIR), on planning CT (pCT) images registered with DP-PET/CT images using RIR and B-spline-based DIR, respectively. Reference GTVs were independently delineated by all radiation oncologists using radiotherapy position (RP)-PET/CT images. DP- and RP-PET/CT images for 14 patients with NPC were acquired using early and delayed scans, respectively. Dice's similarity coefficient (DSC), mean distance to agreement, and volume agreement with reference GTVs were compared by considering the interobserver variability in reference contours. Results The average DSCs for GTV(RIR) and GTV(DIR) were 0.77 and 0.77, which were acceptable for GTV delineation. There were no statistically significant differences between GTV(RIR) and GTV(DIR) in all evaluation indexes (p > 0.05). Furthermore, the correlation between neck flexion angle differences and GTV accuracy was not statistically significant (p > 0.05). Conclusion RIR was a feasible choice compared with the B-spline-based DIR in GTV delineation for NPC under variations of neck flexion angle.