Uncertainty estimations for quantitative in vivo MRI T1 mapping

Mapping the longitudinal relaxation time (T1) of brain tissue is of great interest for both clinical research and MRI sequence development. For an unambiguous interpretation of in vivo variations in T1 images, it is important to understand the degree of variability that is associated with the quantitative T1 parameter. This paper presents a general framework for estimating the uncertainty in quantitative T1 mapping by combining a slice-shifted multi-slice inversion recovery EPI technique with the statistical wild-bootstrap approach. Both simulations and experimental analyses were performed to validate this novel approach and to evaluate the estimated T1 uncertainty in several brain regions across four healthy volunteers. By estimating the T1 uncertainty, it is shown that the variation in T1 within anatomic regions for similar tissue types is larger than the uncertainty in the measurement. This indicates that heterogeneity of the inspected tissue and/or partial volume effects can be the main determinants for the observed variability in the estimated T1 values. The proposed approach to estimate T1 and its uncertainty without the need for repeated measurements may also prove to be useful for calculating effect sizes that are deemed significant when comparing group differences. ⺠Knowledge of the T1 uncertainty is vital to interpret quantitative T1 maps. ⺠Inversion-recovery MRI data was combined with wild bootstrap analysis. ⺠Thus, the minimum uncertainty was estimated per voxel, without repeated scans.