Multi-temporal fusion of abdominal CT images for effective liver cancer diagnosis

Medical imaging has encountered exuberant developments obligated to its capacity to win data regarding human advantage for the motive of diagnosis. The main goal of liver perfusion imaging is to enliven the legitimacy in the characterization of liver disorders. Integration of information from the acquired multitemporal CT images can be achieved by means of image fusion, which is the process of integrating the relevant evidence from a set of source images directed towards a single composite image, where the resultant conception will be greater informative and meticulous than entire of the input images. To fuse multiple images, there should be one-to-one pixel correspondence which is achieved through image registration. In this paper, an fusion in NSCT domain using Phase Congruency and Log- Gabor fusion rule is proposed. The combination of these two fusion rules can preserve more information from source images and thus improves the quality of the fused image in a substantial manner. The statistical assessment metrics such as Spatial Frequency, SSIM, Cross Correlation, Cross Entropy and PSNR for the proposed framework provides an accurate analysis and comparison is carried out with the existing Wavelet and Contourlet transforms.