Medical image fusion framework for neuro brain analysis

Abstract Multimodal neuro image fusion only partially confines the spatial content. Due to variation in image modalities, dark, less clear, and gloomy fusion results are obtained. To overcome these limitations and improve the results of existing techniques, we propose a two phase fusion method using fractional Fourier transform prefixed with the benchmark technique to fuse anatomical and functional brain images. The fractional Fourier transform of the original source images is computed and subsequently in the second phase benchmark techniques are applied on the processed images. The fusion rules used for the study are weighted energy rules for low frequency components and variance-based fusion rules for the high frequency component obtained from transforms. Anatomical and functional images from four brain analysis cases of cerebrovascular disease, neoplastic disease, degenerative disease and from inflammatory or infectious disease categories are evaluated to exemplify the study. Assessment metrics for objective evaluation and subjective evaluation were performed.