摘要目的 探讨基于肝血管树智能对位方法的超声-CT/MR融合成像技术的可行性与简便性.方法 采用Philips Epiq 7的PercuNav融合成像系统对22例增强CT或MR发现肝内局灶性病变的患者进行超声-CT/MR融合成像检查.分别采用系统内置的基于肝血管树智能对位方法和常规的内定标对位方法作配准融合,对比两种方法的配准成功率、初步配准所需时间、初步配准误差、微调配准次数、微调配准时间及配准所需总时间.结果 基于肝血管树智能对位方法和常规的内定标对位方法的配准成功率分别为72.73% 和95.45%,初步配准所需中位时间分别为16.5 s(10~30 s)和13 s(8~24 s),初步配准中位误差分别为3 mm(1~14 mm)和14 mm(2~43 mm),微调配准中位次数分别为0次(0~2次)和1次(0~3次),微调配准中位时间分别为0 s(0~46 s)和30 s(0~88 s),配准所需总中位时间分别为20 s(12~61 s)和42 s(9~102 s).其中,两种方法的配准成功率差异无统计学意义(P >0.05);常规的内定标对位方法初步配准所需时间优于基于肝血管树智能对位方法(P <0.05);而在初步配准误差、微调配准次数、微调配准时间及配准所需总时间等项目,基于肝血管树智能对位方法结果优于常规的内定标对位方法(P <0.05).结论 基于肝血管树智能对位方法的超声-CT/MR融合成像技术具有较高的可行性;相对于常规的内定标对位方法,其操作更为简便高效.

abstractsObjective To explore the feasibility and convenience of automatic registration ultrasound-CT/MR fusion imaging based on hepatic vessel trees. Methods The PercuNav fusion imaging system of Philips Epiq 7 was used to perform ultrasound-CT/MR fusion imaging on 22 patients with focal liver lesions detected by contrast-enhanced CT or MR.Both automatic registration ultrasound-CT/MR fusion imaging based on hepatic vessel trees and the conventional ultrasound-CT/MR fusion imaging based on internal anatomic landmarks were employed for alignment in these patients.The results including the success rate of registration,duration time of initial registration,error of initial registration,number of times of fine-tuning, duration time of fine-tuning and the overall duration time of registration were compared between these two methods.Results The success rates of registration,duration time of initial registration,errors of initial registration,numbers of times of fine-tuning,duration time of fine-tuning and the overall duration time of registration for automatic registration ultrasound-CT/MR fusion imaging based on hepatic vessel trees and the conventional ultrasound-CT/MR fusion imaging based on internal anatomic landmarks were 72.73% and 95.45%,16.5 s (10~30 s) and 13 s (8~24 s),3 mm (1~14 mm) and 14 mm (2~43 mm),0 time (0 to 2 times) and 1 time (0~3 times),0 s(0~46 s) and 30 s (0~88 s),and 20 s (12~61 s) and 42 s (9~102 s),successively and respectively. There was no statistically significant difference in the success rates between these two methods ( P >0.05).The duration time of initial registration of conventional method was less than that of automatic registration method( P <0.05).The error of initial registration,number of times of fine-tuning,duration time of fine-tuning and the overall duration time of registration of automatic registration method were superior to those of conventional method ( P < 0.05).Conclusions Automatic registration ultrasound-CT/MR fusion imaging based on hepatic vessel trees is feasible. It is also more convenient than conventional ultrasound-CT/MR fusion imaging based on internal anatomic landmarks.