The present study focuses on the design of feature based efficient fusion scheme using multiscale shift invariant shearlet transform. For enhancing the obscured but spectacular detailing of brain MRI which is necessary for analyzing the affected tissues, a flexible approach of vague set (VS) theory based segmentation method has been proposed. The existence of non null hesitation in VS theory makes it appropriate for modeling the high degrees of uncertainties/impreciseness presents in MRI. Hence the proposed segmentation method efficiently captures the salient features and fine structures of MRI and makes the irrelevant artifacts smooth. As the relevant local information related to energy activity level, dominant textural variation, spatial structures belonging to edges/contours, likelihood of neighboring contrast distribution, signal complexity are important to produce the fused image, various indices such as root mean square value of local energy (RMSLE), local information entropy (LIE), local contrast index (LCI) and local standard deviation (LSTD) of subband coefficients are captured by placing a 3 × 3 kernel. In this view, principle component analysis (PCA) approach is addressed to produce a composite subband (CSb) image carrying all those local information. Finally, the fused image is constructed based on the strength of CSb for all source images. Experimental results show that the proposed fusion approach is able to integrate utmost information content of the source images and preserve color saliency very efficiently.