Changes in myelin water fraction (MWF) represent a biomarker for central nervous system disease. However, high quality mapping of MWF is challenging, requiring very high signal-to-noise ratio for accurate and stable results. In this work, we demonstrate the potential of a new multispectral filter to permit high quality MWF mapping using in-vivo GRASE brain imaging datasets. Indeed, unlike conventional averaging filters, our filter permits substantial reduction of the random variation in derived MWF estimates while preserving edges and small structures. Finally, our results regarding patterns of MWF as a function of age are consistent with recent literature.