An uncertainty aided framework for learning based liver T 1ρ mapping and analysis

Objective. Quantitative T-1 rho imaging has potential for assessment of biochemical alterations of liver pathologies. Deep learning methods have been employed to accelerate quantitative T-1 rho imaging. To employ artificial intelligence-based quantitative imaging methods in complicated clinical environment, it is valuable to estimate the uncertainty of the predicated T-1 rho values to provide the confidence level of the quantification results. The uncertainty should also be utilized to aid the post-hoc quantitative analysis and model learning tasks. Approach. To address this need, we propose a parametric map refinement approach for learning-based T-1 rho mapping and train the model in a probabilistic way to model the uncertainty. We also propose to utilize the uncertainty map to spatially weight the training of an improved T-1 rho mapping network to further improve the mapping performance and to remove pixels with unreliable T-1 rho values in the region of interest. The framework was tested on a dataset of 51 patients with different liver fibrosis stages. Main results. Our results indicate that the learning-based map refinement method leads to a relative mapping error of less than 3% and provides uncertainty estimation simultaneously. The estimated uncertainty reflects the actual error level, and it can be used to further reduce relative T-1 rho mapping error to 2.60% as well as removing unreliable pixels in the region of interest effectively. Significance. Our studies demonstrate the proposed approach has potential to provide a learning-based quantitative MRI system for trustworthy T-1 rho mapping of the liver.

Automated summary

This study proposes a parametric map refinement approach for learning-based T-1 rho mapping. By training the model probabilistically to model uncertainty and utilizing uncertainty maps for weighted training of an improved mapping network, the mapping performance is enhanced and unreliable values are effectively removed. The results demonstrate the potential of this method to provide a trustworthy learning-based quantitative MRI system for T-1 rho mapping of the liver.