论文部分内容阅读
血管内超声(IVUS)图像的仿真有助于检验诸如图像分割等图像处理算法的性能。提出一种IVUS图像仿真的方法。该方法在极坐标图像生成模型中引入环晕、导丝伪影,并分别对粥样硬化斑块的纤维、脂质、钙化三个区域进行模拟,实现静态图像的仿真;运用血管随心脏搏动的变化规律,实现序列图像的仿真。通过对15例真实IVUS图像、每例各50次的实验表明,相对于传统的极坐标图像生成模型,该方法仿真的图像与真实图像的相关系数提高了56.9%,互信息提高了24.3%,其仿真效果更加接近真实图像。
Simulation of intravascular ultrasound (IVUS) images helps test the performance of image processing algorithms such as image segmentation. A method of IVUS image simulation is proposed. The method introduces halo and guidewire artifact into the polar image generation model and simulates the three regions of atherosclerotic plaque in fiber, lipid and calcification respectively to realize the static image simulation. The change rule, realizes the simulation of sequence image. The experiments of 15 real IVUS images and 50 examples of each showed that the correlation coefficient between the simulated image and the real image increased by 56.9% and the mutual information increased by 24.3% compared with the traditional polar image generation model. Its simulation effect is closer to the real image.