基于Shearlet变换和TV模型的医学图像融合

Medical Image Fusion Based on Shearlet Transform and Total Variation Model

为了综合多模态医学图像的互补信息,为临床诊断和辅助治疗提供更为充分有效的依据,提出了一种基于Shearlet变换和全变差(Total variation,TV)模型的含噪医学图像融合方法。首先对源图像(CT/MRI图像或CT/PET图像)进行Shearlet变换,产生一个低频子带和若干高频子带。然后对低频子带采用基于区域方差的融合策略,以完整地保留源图像的解剖信息或功能信息;对于高频子带,则利用TV模型进行去噪预处理,避免噪声对融合结果的干扰,再采用改进拉普拉斯能量和(Sum-modified-Laplacian,SML)的融合策略,最大程度地融合边缘、细节信息。大量实验结果表明,与近年来提出的3种融合方法相比,本文提出的方法对无噪声和有噪声的医学图像都能取得更好的融合效果,融合图像完整地保留了源图像的信息,目标的边缘、细节清晰,计算效率也有所提高。

A fusion method for noisy medical images based on shearlet transform and total variation model is proposed.The purpose of the proposed method is to integrate complementary information of different modal medical images,which can provide clinical diagnosis and adjunctive therapy with plenty efficient basis.Firstly,shearlet transform is performed on images （CT/MRI images or CT/PET images）.A low-frequency sub-band and several high-frequency sub-bands are produced in each image.Then the fusion rule based on region variance is adopted for the low-frequency sub-band,which can fully preserve the anatomical or functional information of the source images.For the high-frequency sub-bands,total variation model is used to suppress noise to avoid the interference with the fusion results.Then the fusion rule based on sum-modified-Laplacian （SML） is adopted to fuse edges and details best.A large number of experimental results show that,compared with three fusion methods in recent years,the proposed method has better fusion performance for both noiseless and noisy medical images.The fused image can fully preserve the information of source images.The edges and details of the targets are clear,while the computational efficiency is also improved.