ADF参数优化方法在MRI图像中的应用

Application of ADF parameter optimization in MRI

作为一个保边去噪的算法,各向异性扩散滤波(anisotropic diffusion filter,ADF)被广泛应用于磁共振成像(magnetic resonance image,MRI)图像的预处理中,且对MRI图像中的莱斯噪声具有很好的去除效果.各向异性扩散滤波参数的选择对于其去噪性能影响很大,为找出滤波器的最佳参数,我们用改进的遗传算法对其进行参数优化,并且采用了一种新的精英选择策略,而且还在交叉和变异过程中采用了自适应的交叉和变异概率,再分别对各向异性扩散滤波的迭代次数t、扩散阈值k以及时间步长λ等三个参数进行选择优化.最后,从峰值信噪比(peak signal-to-noise ratio,PSNR)、结构相似性指数(structural similarity index metric,SSIM)、均方差(mean squared error,MSE)三个方面,将经过参数优化的各向异性扩散滤波器对脑部MRI进行去噪处理,并与其它参数下的滤波结果进行对比.实验结果表明,经过参数优化的各向异性滤波器,无论是从视觉上还是相关评价指标上,均优于其它参数情况下的去噪效果.

As an edge-preserving denoising algorithm,anisotropic diffusion filter（ADF）is widely used in MRI pre-processing and it is very effective to remove the Rice noise in the MRI image.Parameters such as t（iterations）,k（diffusion threshold）andλ（time step）have much influence on the performance of ADF.They were optimized based on an enhanced genetic algorithm（GA）which used a new elite choice strategy and an adaptive probability of crossover and mutation.The results of denoising with proposed parameters and the previously used ADFs such as the peak signal-to-noise ratio（PSNR）,structural similarity（SSIM）and the mean squared error（MSE）were compared.The experimental results showed that the ADF with parameters optimized by GA had better performance than the ADF with other parameters in both the vision and the quality metrics.