Uncertainty-aware self-supervised neural network for liver T 1ρ mapping with relaxation constraint

Abstract

Abstract Objective. T 1 ρ mapping is a promising quantitative MRI technique for the non-invasive assessment of tissue properties. Learning-based approaches can map T 1 ρ from a reduced number of T 1 ρ weighted images but requires significant amounts of high-quality training data. Moreover, existing methods do not provide the confidence level of the T 1 ρ estimation. We aim to develop a learning-based liver T 1 ρ mapping approach that can map T 1 ρ with a reduced number of images and provide uncertainty estimation. Approach. We proposed a self-supervised neural network that learns a T 1 ρ mapping using the relaxation constraint in the learning process. Epistemic uncertainty and aleatoric uncertainty are modelled for the T 1 ρ quantification network to provide a Bayesian confidence estimation of the T 1 ρ mapping. The uncertainty estimation can also regularize the model to prevent it from learning imperfect data. Main results. We conducted experiments on T 1 ρ data collected from 52 patients with non-alcoholic fatty liver disease. The results showed that when only collecting two T 1 ρ -weighted images, our method outperformed the existing methods for T 1 ρ quantification of the liver. Our uncertainty estimation can further regularize the model to improve the performance of the model and it is consistent with the confidence level of liver T 1 ρ values. Significance. Our method demonstrates the potential for accelerating the T 1 ρ mapping of the liver by using a reduced number of images. It simultaneously provides uncertainty of T 1 ρ quantification which is desirable in clinical applications.