An improved combination of T1-weighted and FLAIR contrasts for myelin water imaging

In vivo myelin imaging often requires long and dedicated MRI scanning techniques. Combining conventional T1-weighted and FLAIR images into the T1w/FLAIR-ratio has been proposed as a myelin-sensitive proxy, however its sensitivity is debated. In this study, we explore the sensitivity to the myelin content of an improved combination of T1w and FLAIR contrasts. A training dataset (N = 23) with reference standard myelin water fraction (MWF) values was used to fit both the T1w and FLAIR contrasts to the MWF using a second order polynomial surface fitting method. This combination was evaluated in an independent dataset (N = 30) with reference MWF values relative to the T1w/FLAIR-ratio by partial correlation analysis adjusting for age and sex. Moreover, the performance of the improved combination was demonstrated in a population-based dataset (N = 4652) by assessing its relation with age. The polynomial fit showed a 27% increase in correlation strength with the MWF compared to the standard T1w/FLAIR-ratio (r = .72 vs. r = .56, p < .01) in the white matter. In the external validation set the polynomial fit also increased the correlation strength compared to the T1w/FLAIR-ratio (r=.64 vs. r = .51, p = .07). Furthermore, in the large population-based dataset, the polynomial fit explained 9.5% more variance in a model with age and age squared, compared to the T1w/FLAIR ratio (R² = .48 vs. R² = .43, p < .001). A second-order polynomial combination of T1w and FLAIR images provides an empirically optimized marker with stronger associations with the reference MWF compared to the conventional T1w/FLAIR-ratio, without dedicated time-consuming imaging.