基于超声图像特征参数检测高强度聚焦超声引起的组织损伤阈值  

Detecting threshold of tissue lesion induced by HIFU based on ultrasonic image feature parameter

作　　者：陈华[1] 钱盛友[1] 谭乔来[1] 邹孝[1] 丁亚军[1] 

机构地区：[1]湖南师范大学物理与信息科学学院,湖南长沙410081

出　　处：《中国医学物理学杂志》2016年第11期1144-1148,共5页Chinese Journal of Medical Physics

基　　金：国家自然科学基金(11174077;11474090);湖南省自然科学基金(11JJ3079)

摘　　要：目的:研究在高强度聚焦超声(HIFU)辐射后,超声图像特征与HIFU辐射引起的组织损伤的关系,以此确定判别组织损伤的参数阈值。方法:在不同HIFU剂量情况下,定点辐射新鲜离体猪肉组织,并保存HIFU辐射前后的超声图像,提取其相关系数、减影图像灰度均值和标准差,并结合支持向量机进行辨识,利用正确识别的样本确定判别组织损伤的参数阈值。结果:对于组织损伤阈值,相关系数、灰度均值和标准差分别有各自的阈值区间,且都能较好地区分组织是否损伤。结论:在训练较好的情况下,相关系数、灰度均值和标准差这3个参数都可以用于组织损伤阈值的划分,但相关系数相对于灰度均值和标准差有更好的区分性能。Objective To determine the parameter thresholds for distinguishing tissue lesion by studying on the relationship between ultrasonic image features and tissue lesion induced by high intensity focused ultrasound （HIFU） radiation. Methods Different HIFU doses were used to fixed-point radiate fresh pork tissue in vitro. The ultrasonic images before and after HIFU radiation were collected. The correlation coefficient, gray mean and standard deviation of subtraction image were extracted, which were combined with support vector machine to distinguish tissue lesion. Finally, the correct identification samples were used to determine the parameter thresholds of the discrimination of tissue lesion. Results The correlation coefficient, gray mean and standard deviation had their respective threshold interval for the threshold of tissue lesion, and could accurately distinguish tissue lesion. Conclusion Under better training conditions, the correlation coefficient, gray mean and standard deviation can be used to divide thresholds of tissue lesion. Compared with gray mean and standard deviation, correlation coefficient has better distinguishing performance.

关 键 词：组织损伤 高强度聚焦超声 相关系数 灰度均值 标准差 阈值 

分 类 号：R445.1[医药卫生—影像医学与核医学] O426.9[医药卫生—诊断学]