基于灰度概率分布-交叉熵法的无损测温方法

Noninvasive temperature estimation method based on gray probability distribution-cross entropy

本研究利用高强度聚焦超声(HIFU)对新鲜离体猪肉组织进行辐照,并利用B超获取辐照前后的超声图像。运用灰度概率分布-交叉熵法对图像进行处理,将辐照前后图像的交叉熵作为检测温度的信息参数。通过曲线拟合,发现多组数据的交叉熵与温度都成近似的线性关系,且斜率非常接近;其确定系数大、误差小、温度分辨率高,明显地优于传统的灰度统计法和离散Fourier变换直流分量法,能够为超声治疗中的无损测温提供实时、可靠的依据。

Irradiating fresh pork in vitro tissue with high intensity focused ultrasound（ HIFU）,the ultrasound images before and after irradiation were obtained by using B ultrasound. After these images were processed by using gray probability distribution-cross entropy method,the cross entropy of images before and after irradiation was regarded as detection temperature information parameter. Through curve fitting,it shows that there is an approximate linear relationship between cross-entropy of multiple sets of data and temperature,and these slopes are very close.Because of the superior R-square,less errors and higher resolution of temperature,this method is better than the traditional grayscale statistics and Discrete Fourier Transform Direct Component method. It can provide a synchronous and reliable basis for noninvasive temperature estimation in HIFU therapy.