【作者】 钟珊珊；

【导师】 王智彪； 王华；

【作者基本信息】 重庆医科大学 ， 生物医学工程， 2010， 硕士

【摘要】 研究背景及目的高强度聚焦超声技术(High Intensity Focused Ultrasound,HIFU)是建立在影像学基础上的治疗技术,没有影像学技术的发展就没有HIFU技术的发展。超声成像因其速度快、无辐射、操作简单而成为HIFU技术中最具有潜力的监控手段之一。诊断超声监控作为众多学者的研究热点,其基本原理是对治疗前后焦区处超声背向散射系数发生改变产生的回声增强现象进行监控。但回声增强这一现象通常是由空化气泡的散射引起的,随着时间的推移,增强的灰度值会逐渐减弱甚至消失,且灰度值的大小和实际的组织损伤没有明确的对应关系,也有学者在研究中发现有时虽有凝固性坏死发生而灰度并未增强。本文针对目前B超监测的缺点,提出利用图像处理手段,通过对比治疗前后诊断B超监控图像的相似性程度,以达到准确判断凝固性坏死的发生时间以及范围,从而实现HIFU的精确治疗。研究方法采用JC-200型高强度聚焦超声肿瘤治疗系统在90w功率下以定点点打方式辐照新鲜离体牛肝组织1s、2s、3s直至15s。储存辐照前、辐照后即刻、1min、2min、3min的超声声像图。根据散斑分析原理将采集的超声图像使用Matlab 7.0软件截取感兴趣的区域,选择两种相似性测度算法对图像进行相似性分析,通过判断相似程度大小来反映组织在HIFU辐照下的变化过程。结果(1)将JC-200自带的声像图分析软件Gray Val 1.0得到的判断结果与病理结果相比较,我们得出在本研究中,Gray Val 1.0判断的正确率为66.0%,错误率为34.0%。(2)利用互相关系数相似性测度评价散斑图像凝固性坏死时,发生凝固性坏死而Gray Val 1.0计算的辐照前后声像图靶区灰度差值大于5的图像与辐照前的散斑图像互相关系数绝对值小于0.79;发生凝固性坏死而Gray Val 1.0计算的辐照前后声像图靶区灰度差值小于5的图像与辐照前的散斑图像互相关系数绝对值小于0.85。(3)利用归一化互信息相似性测度判断时,此方法对噪声较敏感,单纯评价两张不同时刻HIFU辐照前的散斑图像I(A,B)均超过0.650,部分样本甚至达到0.700。而随着治疗时间的延长,虽然散斑图已发生了明显改变,归一化互信息值呈缓慢增长趋势,但幅度很小(增大不超过0.100)。综上所述,利用散斑分析原理检测声像图的相似性程度,可以作为一种新的HIFU治疗过程中凝固性坏死的检测方法。但关键之处在于如何选取最合适的相似性测度和感兴趣区域(Region Of Interest,ROI)大小,以实现凝固性坏死的准确检测。更多还原

【Abstract】 Background and PurposeHIFU is based on the imaging techniques, it would never make progress without the development of imaging technology .Ultrasound has been regard as one of the most potential monitored methods as its speed, no radiation,and simple.The substance of ultrasound to monitor is that there will be echo enhancement phenomenon after HIFU treatment caused by the scattering of cavitation bubbles. While,as time goes by this phenomenon will gradually weaken or even disappear, and the gray value does not correspond to the actually size of tissue damage area.In this paper a new method has been proposed according to the shortcomings were discussed,by comparing the similarity of monitoring images before HIFU treatment to after.Research MethodsJC-200 High-intensity focused ultrasound tumor therapy system has been used to radiate in vitro bovine liver with 90 w,time continues from 1 s,2s until 15s. Ultrasonographic of before HIFUradiation and immediately after radiation,1 min after radiation ,2min after radiation,3 min after radiation have been saved.According to the principles of ultrasound speckle analysis Matlab 7.0 has been used to obtain ROI,and then analysised the similarity of different images by two kinds of similarity measures .Results1) Gray Val 1.0 judge the results obtained Compared with the pathological results, in this study, the accuracy of Gray Val 1.0 was 64.0%, the error rate was 34.0%2) Cross correlation similarity has been used as the a measure to evaluate coagulation necrosis .The cross correlation similarity is less than 0.79 when both Gray Val 1.0 Calculation of target gray-scale ultrasonography before and after radiation the difference is greater than 5 and the pathological result is positive..The cross correlation similarity is less than 0.85 when both Gray Val 1.0 Calculation of target gray-scale ultrasonography before and after radiation the difference is less than 5 and the pathological result is positive.3) Normalized mutual information similarity has been used as the a measure to evaluate coagulation necrosis too.This method is more sensitive to noise.Simple evaluation of 2 different time speckle images before HIFU treatment I(A,B) are over 0.650,and even some samples to 0.700.With longer time of treatment,althogh the speckle pattern has significant changes,normalized mutual information values increased slowly (less than 0.100)In summary,speckle analysis would be a new method to detect coagulation necrosis.However how to choose the most suitable similarity measure and the size of ROI is the key way to make success.更多还原