Abstract

Info

Title:: Medical image fusion method based on non-subsampled shearlet transform and pulse coupled neural network

Author(s):: TIAN Juanxiu1; 2; LIU Guocai1; 1. School of Electrical and Information Engineering, Hunan University, Changsha 410082, China; 2. School of Computer and Communication, Hunan Institute of Engineering, Xiangtan 411104, China

Keywords:: ; Keywords: non-subsampled shearlet transform; pulse coupled neural network; medical image fusion; positron emission tomography; computed tomography; magnetic resonance imaging

Abstract:: Abstract: Objective To fuse positron emission tomography/computed tomography/magnetic resonance imaging (PET/CT/MRI) images for providing more information such as edge and texture features in fused images to distinguish lesions and tumors from normal tissues and organs and providing more useful information for diagnosis. Methods A fusion method based on non-subsampled shearlet transform (NSST) and pulse coupled neural network (PCNN) model was proposed. A weighted method based on the total of local regional energy was applied to fuse NSST low-frequency coefficients. Then based on the times of PCNN neuron activation, the NSST high-frequency direction coefficients were selected. Finally, the fused images were obtained by inverse NSST. Results Experiments performed on 7 groups of MRI/PET and CT/PET image sets demonstrated that the image visual effects of the fused images were good and that the proposed method had better performances than other algorithms in the comprehensive assessment of mutual information, edge similarity, gradient similarity and spatial frequency. Conclusion The proposed method can adaptively preserve the edge and textures of the source images and achieve a good fusion performance.