Cram\'er-Rao bound-informed training of neural networks for quantitative MRI

TL;DR

This paper introduces a CRB-informed loss function for training neural networks in quantitative MRI, improving estimation accuracy across heterogeneous parameter spaces by balancing parameter difficulty and providing an absolute performance metric.

Contribution

The paper proposes a theoretically grounded CRB-based loss function for neural network training in quantitative MRI, addressing challenges in heterogeneous parameter spaces and enabling absolute estimator evaluation.

Findings

CRB-based loss improves estimation accuracy in heterogeneous spaces

Networks trained with CRB loss outperform those with MSE loss in experiments

CRB loss provides an absolute measure of estimator efficiency

Abstract

Neural networks are increasingly used to estimate parameters in quantitative MRI, in particular in magnetic resonance fingerprinting. Their advantages over the gold standard non-linear least square fitting are their superior speed and their immunity to the non-convexity of many fitting problems. We find, however, that in heterogeneous parameter spaces, i.e. in spaces in which the variance of the estimated parameters varies considerably, good performance is hard to achieve and requires arduous tweaking of the loss function, hyper parameters, and the distribution of the training data in parameter space. Here, we address these issues with a theoretically well-founded loss function: the Cram\'er-Rao bound (CRB) provides a theoretical lower bound for the variance of an unbiased estimator and we propose to normalize the squared error with respective CRB. With this normalization, we balance the contributions of hard-to-estimate and not-so-hard-to-estimate parameters and areas in parameter space, and avoid a dominance of the former in the overall training loss. Further, the CRB-based loss function equals one for a maximally-efficient unbiased estimator, which we consider the ideal estimator. Hence, the proposed CRB-based loss function provides an absolute evaluation metric. We compare a network trained with the CRB-based loss with a network trained with the commonly used means squared error loss and demonstrate the advantages of the former in numerical, phantom, and in vivo experiments.