基于超声图像小波分形维和熵监测生物组织凝固性坏死

Monitoring Coagulation Necrosis of Biological Tissue Based on Wavelet Fractal Dimension and Entropy

高强度聚焦超声(HIFU)治疗肿瘤过程中,辨别组织是否发生凝固性坏死非常重要.本文从数字图像处理的角度出发,研究利用超声图像的小波分形维和熵监测生物组织变性的可行性.将高强度聚焦超声打击新鲜猪肉组织后获得的超声图像与打击前的超声图像做差值处理,截取焦域中心64×64像素的图像,然后提取其小波分形维和熵作为参量信息,比较不同参量对组织凝固性坏死的辨识效果.实验表明,提取小波分形维和熵特征并输入BP神经网络,相对于仅使用某种特征参数而言有更高的识别率,可为临床监测生物组织是否有凝固性坏死发生提供有力的依据.

It is very important to recognize whether the tissue was coagulation necrosis or not in the process of High Intensity Focused Ultrasound（HIFU）treatment of tumor.From the view of digital image processing,the feasibility of using wavelet fractal dimension and entropy of ultrasonic image to monitor the biological tissue degeneration is studied.Firstly,the subtraction image of 64＊64pixel in focal region center was obtained from subtracting the ultrasonic image before High Intensity Focused Ultrasound（HIFU）And after High Intensity Focused Ultrasound（HIFU）against fresh pork tissue.And the wavelet fractal dimension and entropy were extracted from it,which were used as the parameter information,then comparing the recognition effect on tissue coagulation necrosis for different parameters.The result shows that combining the wavelet fractal dimension with entropy to construct BP neural network can get a higher recognition rate compared to use one characteristic parameter.It can provide a strong basis for the clinical monitoring whether the biological tissue has the coagulation necrosis or not.