US-MRI配准融合技术在超声消融子宫肌瘤中精确影像识别及智能化培训的研究

发布时间：2018-05-26 02:07

  本文选题：HIFU + 子宫肌瘤　； 参考：《重庆医科大学》2017年硕士论文

【摘要】：子宫肌瘤是育龄期妇女中极为常见的良性肿瘤,该肿瘤会导致一系列健康问题。高强度聚焦超声(High Intensity Focused Ultrasound,HIFU)消融子宫肌瘤是一种无创治疗方法,其凭借微创性和保留患者生育能力等特性,正越来越受到妇科医生的重视。随着超声引导的HIFU技术广泛应用于临床,突破超声图像(Ultrasound Image,US)质量差、噪声大、视野窄的局限性,将超声图像与磁共振图像(Ultrasound Image and Magnetic resonance image,US-MRI)配准、融合技术用于提高超声消融子宫肌瘤中的影像识别精确度和推进智能化培训成为新的紧迫需求。为此超声医疗国家工程研究中心HIFU研究团队基于US-MRI配准、融合技术、三维重建技术等智能技术,开发出HIFU虚拟现实系统。本研究旨在评估超声消融子宫肌瘤中US-MRI配准、融合技术的精确影像识别能力和基于US-MRI配准技术的虚拟现实培训的效果,以期为其在超声消融子宫肌瘤中的运用、推广提供依据。目的:1.探讨US-MRI图像配准、融合技术在超声消融子宫肌瘤中的标定结构及其可行性。2.测量超声消融子宫肌瘤术中US-MRI图像配准技术的配准精确度。3.评估HIFU虚拟现实系统辅助超声消融子宫肌瘤手术定位培训的效果。方法:1.前瞻性地评估78例子宫肌瘤患者78组图像的配准与融合效果,筛选配准标定结构,评价配准技术的性能指标:配准精确度、可靠性和临床可行性。2.根据“客观结构化临床考试”规则设计虚拟测试项目,从信度和效度上评估虚拟培训及测试项目的质量。3.2015年12月到2017年1月在重庆医科大学附属第一医院超声消融治疗中心(卫生部高强度聚焦超声肿瘤治疗培训基地)参加超声消融子宫肌瘤手术培训的学员8名,学习内容为HIFU治疗前预定位。学员分配到虚拟现实培训组(虚拟组)和传统培训组(传统组),每组4人。同一名高级医生选出适合学员定位的患者240例,通过双盲法(学员和患者互相不能选择)分配给每位学员30例患者,按医生定位的顺序分为6期,每期每位医生完成5例患者的治疗前预定位。以定位时间为指标评价学习曲线,比较HIFU虚拟现实系统辅助超声消融子宫肌瘤手术定位培训的效果。结果:1.图像配准标定结构,78组图像,76组图像可以实现图像配准操作,2组子宫前位的图像(因为US与MRI显示的子宫位置变化过大不能实现图像配准)。76组图像配准后,清晰度均有提高。子宫、子宫肌瘤、耻骨、骶尾骨都可以根据边界线、标定点进行US与MRI的配准。用子宫的边界作为标志来配准同一子宫平面,配准清晰度显著增加(P=0.000);用子宫肌瘤的边界作为标志配准同一子宫肌瘤平面,配准清晰度显著增加(P=0.000);用耻骨的边界作为标志配准同一耻骨平面,配准清晰度显著增加(P=0.000);用骶岬作为标志点配准骶尾骨,配准清晰度显著增加(P=0.000)。2.配准的可行性评价结果,效果A,图像配准后清晰度有很好的改善:US与MRI有很好的相关性,扩大了视野,提高了病灶边界的清晰度。所占比例82.05%。效果B,图像配准后清晰度没有改善:US与MRI没有相关性:不能通过配准图像进行诊断。所占比例2.56%。效果C,在US上已经很清楚的,不需要通过配准图像进行诊断。所占比例为15.38%。3.配准精确度,子宫图像中,76组可实现图像配准操作,故以子宫的边界为配准标志物,配准子宫的配准误差为3.20±0.10mm,以子宫肌瘤的边界为配准标志物,配准子宫肌瘤的配准误差为3.00±1.25mm。45例患者的内膜边界在US上可见,故以内膜的边界为配准标志物,配准内膜的配准误差为2.20±0.45mm;33例患者的骶岬在US上可见,故以骶岬作为配准标志物,配准骶尾骨的配准误差为1.82±0.48mm;22例患者的耻骨边界在US上可见,故以耻骨边界为配准标志物,配准耻骨的配准误差为1.64±0.32mm。4.图像融合效果评价,以子宫肌瘤的边界作为标志配准图像后,实现MRI与US的全局融合。70组融合图像,3组(4.29%)融合效果为C级(C.融合图像的清晰度与配准后的图像相比病灶边界及周边解剖结构没有改善),67组(95.71%)融合效果为D级(D.融合图像的清晰度与配准后的图像相比病灶边界及周边解剖结构的清晰度下降)5.定位能力测试结果,可信度,Cronbach'sα=0.90。定位能力测试总成绩,虚拟组得分40.00±1.63分,传统组得分26.75±5.74分。虚拟组得分显著高于传统组得分(t=4.421,P=0.004)。定位能力测试子项目:(1)定位的设备操作能力测试项目,可信度Cronbach’sα=0.91,测试成绩,虚拟组得分16.25±2.36分,传统组得分12.00±4.24分。虚拟组得分与传统组得分相比差异无统计学意义(t=1.753,P=0.131);(2)定位中标志性解剖结构识别能力测试,可信度Cronbach’sα=0.87,测试成绩,虚拟组得分14.00±1.41分,传统组得分10.50±1.92分。虚拟组得分显著高于传统组得分(t=2.942,P=0.026);(3)定位中疑难情况判别能力测试项目,可信度Cronbach’sα=0.88,测试成绩,虚拟组得分9.75±2.63分,传统组得分4.25±1.25分。虚拟组得分显著高于传统组得分(t=3.774,P=0.017)。6.学习曲线分析,虚拟现实技术培训组6期的学习曲线和传统培训组6期的学习曲线分析结果:在定位操作实践的1至6期,虚拟现实技术培训组的学习曲线处于传统培训组的学习曲线下方;其中定位操作实践的1至4期,虚拟现实技术培训组的学习曲线与传统培训组的学习曲线的间距呈现振荡渐近状态,从4期以后,虚拟现实技术培训组的学习曲线与传统培训组的学习曲线呈现平稳渐近趋势。结论:1.在超声消融子宫肌瘤中US-MRI的配准、融合是可以实现的。US-MRI配准技术可以在一定程度上,弥补MRI不能实时监控的缺点和US分辨力低的缺点,提高图像的清晰度、扩大视野、辅助医生了解US上被伪影挡住的图像信息,有效提高超声消融子宫肌瘤中的影像识别效果。2.本研究的评估结果显示虚拟现实考核方案的可信性度、普适性好。虚拟现实系统辅助培训与传统的培训相比在学员的初始学习阶段具有更好的培训效果。
[Abstract]:Uterine myoma is a very common benign tumor of women of childbearing age. This tumor can lead to a series of health problems. High Intensity Focused Ultrasound (HIFU) ablation of uterine myoma is a noninvasive treatment. It is becoming more and more important for gynecologists with the characteristics of minimally invasive and retention of the patient's fertility. As the ultrasound guided HIFU technology is widely used in clinic, it breaks through the limitations of the poor quality of Ultrasound Image (US), large noise and narrow vision, and registration of ultrasonic images with magnetic resonance images (Ultrasound Image and Magnetic resonance image, US-MRI). Fusion technique is used to improve the image recognition in ultrasonic ablation of uterine myoma. The HIFU research team of the ultrasound medical National Engineering Research Center has developed a HIFU virtual reality system based on US-MRI registration, fusion technology and 3D reconstruction technology. This study aims to evaluate the accuracy of US-MRI registration and fusion technology in ultrasonic ablation of uterine myoma. Image recognition ability and the effect of virtual reality training based on US-MRI registration technology in order to provide the basis for its application in ultrasonic ablation of uterine myoma. Objective: 1. to explore the calibration structure and feasibility of US-MRI image registration, fusion technique in ultrasonic ablation of uterine myoma, and to measure the US-MRI map in the operation of ultrasonic ablation of uterine myoma by ultrasonic ablation. Image registration accuracy.3. evaluated the effect of HIFU virtual reality system assisted ultrasound ablation of uterine myoma operation location training. Method: 1. the registration and fusion effect of 78 groups of images in 78 cases of hysteromyoma were evaluated prospectively, the registration and calibration structure was screened, and the performance indexes of registration techniques were evaluated: accuracy, reliability and presence of registration. Bed feasibility.2. design virtual test project based on the "objective structured clinical examination" rules, evaluate the quality of virtual training and testing project from reliability and validity from December to January 2017 in First Affiliated Hospital of Chongqing Medical University ultrasound ablation treatment center (Department of high intensity focused ultrasound tumor treatment training base of the Ministry of health). 8 students were trained with ultrasonic ablation of uterine myoma. The learning content was pre positioning before HIFU treatment. Students were assigned to the virtual reality training group (virtual group) and the traditional training group (traditional group), with 4 people in each group. The same senior doctor selected 240 patients suitable for the students to locate the students and assigned them to each other by double blind method (students and patients could not choose from each other). 30 patients were divided into 6 stages according to the doctor's positioning order. Each doctor completed the pre location of 5 patients before the treatment. The learning curve was evaluated with the location time as the index, and the effect of the HIFU virtual reality system assisted ultrasound ablation of hysteromyoma was compared. The results were: 1. image registration calibration structure, 78 groups of images, 76 group drawings. Image registration operations can be achieved. 2 groups of images of the anterior uterine position (because of the large changes in the position of the uterus that US and MRI show too large can not achieve image registration) the definition of the.76 group is improved. The uterus, the uterus myoma, the pubic bone, the sacral tail can all be registered with the US and MRI according to the boundary line, and the boundary of the uterus is used as the mark. The registration clarity of the same uterine plane was significantly increased (P=0.000). The registration definition of the same uterus myoma was registered with the boundary of uterine myoma as a sign, and the registration clarity increased significantly (P=0.000). The registration of the same pubic plane with the boundary of the pubis was used as a sign, and the registration clarity increased significantly (P=0.000); sacral headland was used as a sign to registration sacrum. The registration clarity increased significantly (P=0.000).2. registration feasibility evaluation results, the effect was A, the clarity of the image registration after registration was well improved: US and MRI have good correlation, expanded the vision and improved the definition of the focus of the focus. The proportion of 82.05%. effect B, the image registration clarity did not improve: US and MRI have no correlation: no correlation: no It can be diagnosed by registration image. The proportion of 2.56%. is C, which is clear on US. It does not need to be diagnosed by registration image. The proportion is 15.38%.3. registration accuracy. In the uterus image, 76 groups can realize image registration operation, so the uterus boundary is used as the matching mark and registration error of the uterus is 3.20 + 0.10mm. The registration error of uterine myoma is a registration marker. The registration error of the registration of uterine myoma is 3 1.25mm.45 cases, so the inner boundary of the endometrium is visible on the US. Therefore, the boundary of the endometrium is the registration marker, the registration error of the endometrium is 2.20 0.45mm; the sacral headland of the 33 patients can be seen on the US, so the sacral headland is used as the registration marker and registration of the sacral tail. The registration error of bone is 1.82 + 0.48mm, and the pubic boundary of 22 cases is visible on US. Therefore, the registration mark of the pubis is the registration mark of the pubis, the registration error of the pubic bone is 1.64 + 0.32mm.4. image fusion. After the registration image of the uterus myoma, the fusion image of the global fusion.70 group of MRI and US is realized, and the fusion of the 3 groups (4.29%) is fused. The effect was C (the definition of the C. fusion image and the image after registration did not improve), and the 67 group (95.71%) was D level (the clarity of the D. fusion image was compared with the image after registration and the clarity of the lesion boundary and the peripheral anatomical structure decreased). The results of the 5. location test, the reliability, and the Cronbach's The total score of the =0.90. localization ability test, the virtual group score 40 + 1.63 points, the traditional group score 26.75 + 5.74 points. The virtual group score was significantly higher than the traditional group score (t=4.421, P=0.004). The positioning ability test sub project: (1) the positioning equipment operation ability test project, the reliability Cronbach 's alpha =0.91, the test result and the virtual group score 16.25 + 2.36 points, The score of the traditional group was 12 + 4.24 points. There was no significant difference in the score of the virtual group compared with the traditional group (t=1.753, P=0.131); (2) the test of the identification ability of the marker anatomy, the reliability of Cronbach 's a =0.87, the score of the test, the score of the virtual group 14 + 1.41, the score of the traditional group 10.50 + 1.92. The score of the virtual group was significantly higher than that of the traditional group. Group score (t=2.942, P=0.026); (3) the test project of difficult situation discriminant ability, reliability Cronbach 's alpha =0.88, test score, 9.75 + 2.63 score of virtual group score and 4.25 + 1.25 score of traditional group. The score of virtual group was significantly higher than that of the traditional group score (t=3.774, P= 0.017).6. learning curve analysis, and 6 period of virtual reality technical training group. Curve and the learning curve analysis of the 6 phase of the traditional training group: the learning curve of the VR training group is under the learning curve of the traditional training group in the 1 to 6 period of the positioning operation practice; in the 1 to 4 period of the positioning operation practice, the distance between the learning curve of the VR training group and the learning curve of the traditional training group is presented. The oscillation is asymptotically asymptotically. From the 4 phase, the learning curve of the virtual reality technology training group and the learning curve of the traditional training group are asymptotically asymptotically asymptotically. Conclusion: 1. the registration of US-MRI in the ultrasonic ablation uterine myoma and the integration of the.US-MRI registration technology can make up for the shortcomings of the MRI cannot be monitored in real time and U. S has the disadvantages of low resolution, improving the clarity of the image, expanding the vision, assisting the doctor to understand the image information that is blocked by the artifact on the US, effectively improving the image recognition effect in the ultrasonic ablation of the uterine myoma. The evaluation results of this study show that the credibility of the virtual reality assessment scheme is good. The virtual reality system assists training and transmission. Unified training has better training effect than the initial learning stage.
【学位授予单位】：重庆医科大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R737.33
，

本文编号：1935532