Multimodal medical image fusion based on intuitionistic fuzzy sets and weighted local energy in nsst domain

Abstract

In the extraction of information from multimodality images, anatomical and functional image fusion became an effective tool in the applications of clinical imaging. Objective: A new approach to fuse anatomical and functional images that use the combination of activity measure and intuitionistic fuzzy sets in NSST domain is presented. First, the high and low-frequency sub-images of source images are obtained by utilizing NSST decomposition, which represents them in multi-scale and multi-directions. Next, the high-frequency sub-images are applied to intuitionistic fuzzy sets, in which the fused coefficients are selected using an activity measure called fuzzy entropy. The multiplication of weighted local energy and weighted sum modified Laplacian is used as an activity measure to fuse the low-frequency sub-images. At last, the reconstruction of the final fused image is done by applying the inverse NSST on the above-fused coefficients. The efficacy of the proposed fuzzy-based method is verifiable by five different modalities of anatomical and functional images. Both subjective and objective calculations showed better results than existing methods.