Mutual-Information Based Optimal Experimental Design for Hyperpolarized $^{13}$C-Pyruvate MRI

TL;DR

This paper introduces an information-theory-based optimal experimental design for hyperpolarized $^{13}$C-Pyruvate MRI, optimizing flip angles to improve the accuracy and precision of tumor metabolism rate measurements, with considerations for spatial variation.

Contribution

It develops a mutual information-based optimization method for experimental design in HP-MRI, incorporating a high-fidelity spatial model and analyzing flip angle schemes for better parameter estimation.

Findings

Time-varying flip angles improve mutual information.

Constant flip angles yield better accuracy with noisy data.

Optimal flip angles identified as 35° for pyruvate and 28° for lactate.

Abstract

A key parameter of interest recovered from hyperpolarized (HP) MRI measurements is the apparent pyruvate-to-lactate exchange rate, $k_{PL}$, for measuring tumor metabolism. This manuscript presents an information-theory-based optimal experimental design (OED) approach that minimizes the uncertainty in the rate parameter, $k_{PL}$, recovered from HP-MRI measurements. Mutual information (MI) is employed to measure the information content of the HP measurements with respect to the first-order exchange kinetics of the pyruvate conversion to lactate. Flip angles of the pulse sequence acquisition are optimized with respect to the mutual information. Further, a spatially varying model (high-fidelity) based on the Block-Torrey equations is proposed and utilized as a control. A time-varying flip angle scheme leads to a higher parameter optimization that can further improve the quantitative value of mutual information over a constant flip angle scheme. However, the constant flip angle scheme leads to the best accuracy and precision when considering inference from noise-corrupted data. For the particular MRI data examined here, pyruvate and lactate flip angles of 35 and 28 degrees, respectively, were the best choice in terms of accuracy and precision of the parameter recovery. Moreover, the recovery of rate parameter $k_{PL}$ from the data generated from the high-fidelity model highlights the influence of diffusion and strength of vascular source on the recovered rate parameter. Since the existing pharmacokinetic models for HP-MRI do not account for spatial variation, the optimized design parameters may not be fully optimal in a more general 3D setting.