A Survey and Analysis of Pixel Level Multisensor Medical Image Fusion using Discrete Wavelet Transform

Abstract

Medical image fusion has become an active research topic due to advances in sensor technology, microelectronics, processing techniques that combine information from different sensors into a single composite image for analysis, interrelation and better clinical diagnosis in much rapid and accurate way. The main objective of medical image fusion using wavelets is to create new image by regrouping the complementary information of multi sensor output.Various medical imaging modalities such as X-ray, Ultrasonics, Computed tomography (CT) magnetic Resonance Imaging (MRI), Positron Emission Tomography (PET), Single Photon Emission tomography (SPECT). This work presents the survey of existing fusion schemes and a novel approach of medical image fusion using Discrete wavelet transform (DWT). The challenge is thus to fuse multi sensor medical images with spectral aspects of low resolution and spatial aspects of a high-resolution image, with out introducing artifacts. In this research multilayer data sets are also tested apart from CT & MRI images. The DWT is applied independently to obtain a set of approximation coefficients, horizontal coefficients, vertical coefficients, and diagonal coefficients for two images. The different independent bands are fused to get the combined output. By applying the Inverse Discrete wavelet transform (IDWT)the fused image is reconstructed. The simulation of fused image when consulted with radiologist and physician, proved better clarity and have high information compared to the first two sensor images. Hence the clinical diagnosis is much faster and accurate.