Detail-Enhanced Medical Image Fusion in NSCT Domain

Multimodal medical image fusion technique plays an important role in clinical applications, such as pathologic diagnosis and surgical options. However, many traditional fusion methods cannot well preserve details of source images in the fused image. To address this problem, a detail-enhanced image fusion scheme based on nonsubsampled contourlet transform (NSCT) and gain control (i.e., NCGC) is developed in this paper, which can effectively combine the spectral information and the spatial features of source images. The proposed method applies power law transformation to tune coefficients of each decomposed subband, and adjusts the strength of subband signals by smooth gain control. Eventually, the fused image with more anatomical details and functional information is constructed by the inverse NSCT. Three pairs of medical images with different modalities and three fusion metrics are applied to validate the feasibility of this algorithm. Experimental results demonstrate that the proposed method can achieve superior performance in both visual perception and objective assessment.