Predicting Uncertainty of Metabolite Quantification in Magnetic Resonance Spectroscopy with Applications for Adaptive Ensembling

Current deep learning methods for metabolite quantification in magnetic resonance spectroscopy do not offer reliable measures for uncertainty. Having a widely applicable measure can aid with the identification of fitting errors and enable uncertainty-based adaptive ensembling of model-based quantification and neural network predictions. In this abstract, we propose a training strategy based on a log-likelihood cost that allows joint optimization of concentration and uncertainty estimation for each individual metabolite. We show that the predicted uncertainties correlate well with the actual estimation errors and that uncertainty-based adaptive ensembling outperforms the individual estimators as well as standard ensembling.