Discrete wavelet transform based principal component averaging fusion for medical images

Image fusion schemes play a vital role in medical image analysis and treatment planning. In this paper, a novel principal component averaging fusion based on discrete wavelet transform is proposed for fusion of CT-MRI and MRI images. Even though multiscale fusion methods result in effective integration of image details; Pixel level fusion methods based on principal component analysis (PCA) schemes do not loose popularity because of conceptual simplicity. Conventional principal components analysis fusion evaluates principal components based on eigen values of the source images. Using discrete wavelet transform (DWT), source images are decomposed into multiscale inputs and the principal components are evaluated for multiscale coefficients. Average of principal components of all these relevant decomposed elements will constitute weights for fusion rule. This method incorporates the advantage of wavelet transform into PCA fusion in the form of eigen values of multiscale representations. Performance of the proposed method is experimented on CT-MRI and MRI medical images. Analysis of qualitative and quantitative metrics clearly demonstrates that this method exhibits superior results than many other well known fusion schemes.