Subject-specific head gradient PNS prediction using only demographics and head dimensions

Motivation: Given the ~3-fold population variability in PNS thresholds, rapid and accurate prediction of PNS thresholds for individual subjects would be valuable. Goal(s): To test the hypothesis that individual PNS thresholds can be predicted using only demographics and head dimensions. Approach: PNS thresholds and localizer images were measured for 30 participants in a head gradient. Demographic (age, sex, BMI) and anatomical (head dimensions and offset from isocenter) variables were used as explanatory variables in multivariate models to predict subject-specific PNS thresholds for each of the three Impulse gradient axes. Results: Subject-specific models succeeded in explaining a majority of the population variance in PNS thresholds. Impact: Multivariate linear models using only select demographics and head dimensions as explanatory variables can estimate subject-specific PNS parameters with reasonable accuracy, explaining ~50-80% of the population variance and permitting the tailoring and tightening of on-scanner PNS limits to the individual.