基于肝表面智能自动对位法的超声-CT/MR融合成像的可行性研究  

Feasibility Study of Intelligent Automatic Registration Ultrasound-CT/MR Fusion Imaging Based on Liver Surface

作　　者：曾庆劲[1] 颜荣华[2] 吴宇轩[1] 龙颖琳[1] 罗丽萍 李凯[1] 许尔蛟[1] 郑荣琴[1] ZENG Qing-jing;YAN Rong-hua;WU Yu-xuan;LONG Ying-lin;LUO Li-ping;LI Kai;XU Er-jiao;ZHENG Rong-qin(Department of Medical Ultrasonics,The Third Affiliated Hospital of Sun Yat-sen University,Guangzhou 510630,China;Department of Radiology,The Third Affiliated Hospital of Sun Yat-sen University,Guangzhou 510630,China)

机构地区：[1]中山大学附属第三医院超声科,广东广州510630 [2]中山大学附属第三医院放射科,广东广州510630

出　　处：《中山大学学报（医学版）》2019年第1期117-123,共7页Journal of Sun Yat-Sen University:Medical Sciences

基　　金：国家重点研发计划项目(2017YFC0112000);国家自然科学基金重点项目(81430038);国家自然科学基金(81401434);广东省科技计划项目(2017A020215082;2017A020215137);广州市科技计划项目(201704020164)

摘　　要：【目的】探讨基于肝表面智能自动对位法的超声-CT/MR融合成像在肝脏局灶性病变定位的可行性及简便性。【方法】采用飞利浦EPIQ7的PercuNav融合成像系统对30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。同时采用系统配置的基于肝表面智能自动对位法(智能法)和常规内定标人工对位法(人工法)进行融合成像配准,比较两种方法的配准成功率、初步配准误差、微调配准次数。【结果】所有30例患者中,智能法和人工法的配准成功率均为96.67%(29/30),且初步配准误差和微调配准次数两种方法之间比较无统计学差异(P>0.05)。进一步分层分析,左肝病例中,人工法的初步配准误差优于智能法,差异有统计学意义(P=0.00),而微调配准次数虽然人工法少于智能法,但两者比较差异无统计学意义(P=0.09);而在右肝病例中,智能法的初步配准误差和微调配准次数均优于人工法,差异有统计学意义(P<0.05)。【结论】基于肝表面智能自动对位法的超声-CT/MR融合成像技术具有较高的配准成功率,可行性较好;相对于常规的内定标人工对位法,对于右肝病灶其融合成像操作过程更为简便高效,有助于降低对操作者融合成像技术经验的依赖性。【Objective】To explore the feasibility and simplicity of intelligent automatic registration ultrasound-CT/MR fusion imaging based on liver surface in localization of focal liver lesions.【Methods】Thirty patients with detected focal liver lesions by contrast-enhanced CT or MR were enrolled for ultrasound-CT/MR fusion imaging using the PercuNav fusion imaging system in PHILPS EPIQ7.Both intelligent automatic registration ultrasound-CT/MR fusion imaging based on liver surface(intelligent method)and the conventional internal plane method(manual method)were used for ultrasound-CT/MR fusion imaging.The success rate of registration,the initial registration error and the times of fine-tuning were compared between these two methods.【Results】In all 30 patients,the success rates of registration were both 96.67%(29/30)using intelligent method and manual method.There was no significant difference between these two methods when compared the initial registration error and the times of fine-tuning(P >0.05).According to the further stratified analysis,in 10 lesions in the left liver,the initial registration error of the manual method was less than that of intelligent method,the difference was statistically significant(P=0.00).Although the times of fine-tuning of the manual method was less than that of intelligent method,there was no statistically significant difference(P=0.09);In 20 lesions in the right liver,the initial registration error and the times of fine-tuning of the intelligent method were superior to those of the manual method.The differences were statistically significant(P<0.05).【Conclusion】Intelligent automatic registration ultrasound-CT/MR fusion imaging based on liver surface is a feasible method with high success rate for ultrasound-CT/MR fusion imaging.Compared with the conventional internal plane method,the fusion imaging process is more simple and efficient for the lesions in right liver.It helps to reduce experience dependence of fusion imaging for the operators.

关 键 词：肝脏 融合成像 智能对位 超声 肝表面 

分 类 号：R735.7[医药卫生—肿瘤]