In this paper, new fusion model of EEG-fMRI is proposed, where the EEG correlation is developed with functional activity of the brain imaging (fMRI) to derive features. The fMRI slices and the EEG signals are correlated using the spectral correlation of selective spectral bands and the MRI spectral density in EEG-fMRI sample. Previously the mapping approach of peak spike active of EEG signal over large spectral density of MRI sample is assumed to beprocessed under no-artefact scenario. Secondly a redundant time feature magnitude is not effective in deriving the epileptic diagnosis of EEG signalling. Both challenges are addressed by our methodology in which feature reduction and feature mapping of selective spectral bands of EEG and MRI spectral density is proposed taking into considerationartefacts present in EEG signal. The proposed methodologies are adopted for the detection of epilepsy disorder. Experimentations are conducted in the openly available fMRI data of six subjects and active database, ACTIVE project at TelAviv Sourasky Medical Center (TASMC) Israel and comparisons are made with existing mapping models. The comparative results and analysis demonstrate the superiority and the reliability of the proposed methodology.