Ervural S.Ceylan M.2020-03-262020-03-2620189.78154E+12https://dx.doi.org/10.1109/SIU.2018.8404559https://hdl.handle.net/20.500.12395/37153Aselsan;et al.;Huawei;IEEE Signal Processing Society;IEEE Turkey Section;Netas26th IEEE Signal Processing and Communications Applications Conference, SIU 2018 -- 2 May 2018 through 5 May 2018 -- 137780In this study, it is aimed to analyze the magnetic resonance (MR) images used in the diagnosis of liver focal lesions using image fusion methods and to help diagnosis by adding automatic segmentation results to the manual segmentation process preferred by experts. For this aim fusions of liver MR images, segmented by a fuzzy method and segmented manually. 120 T1-weighted dynamic contrast-enhanced liver MR images of pre-contrast phase, arterial phase, portal vein phase and late venous phase, taken from 30 different patients, were used. Each phase image is also fused with images segmented by the fuzzy c-means algorithm in the same phase, so that the lesion surfaces and contours are displayed on the segmented image manually. Thus, the significance of the lesion was increased before the information in the MR image in which the liver function information was displayed was lost. The resulting new image contains more useful information for automatic decision systems. The results obtained were evaluated using structural similarity index, peak signal-to-noise ratio and fusion factor quality metrics. © 2018 IEEE.tr10.1109/SIU.2018.8404559info:eu-repo/semantics/closedAccessDiscrete wavelet transformFuzzy c-meansImage fusionSegmentationConference Object14N/A