An efficient approach to optimal experimental design for magnetic resonance fingerprinting with B-splines

Purpose: To introduce a computationally efficient approach to optimizing the data acquisition parameters of MR Fingerprinting experiments with the Cramer-Rao bound. Methods: This paper presents a new approach to the optimal experimental design (OED) problem for MR Fingerprinting, which leverages an early observation that the optimized data acquisition parameters of MR Fingerprinting experiments are highly structured. Specifically, the proposed approach captures the desired structure by representing the sequences of data acquisition parameters with a special class of piecewise polynomials known as B-splines. This incorporates low- dimensional spline subspace constraints into the OED problem, which significantly reduces the search space of the problem, thereby improving the computational efficiency. With the rich B-spline representations, the proposed approach also allows for incorporating prior knowledge on the structure of different acquisition parameters, which facilitates the experimental design. Results: The effectiveness of the proposed approach was evaluated using numerical simulations, phantom experiments, and in vivo experiments. The proposed approach achieves a two-order-of-magnitude improvement of the computational efficiency over the state-of-the-art approaches, while providing a comparable signal-to-noise ratio efficiency benefit. It enables an optimal experimental design problem for MR Fingerprinting with a typical acquisition length to be solved in approximately 1 min. Conclusions: The proposed approach significantly improves the computational efficiency of the optimal experimental design for MR Fingerprinting, which enhances its practical utility for a variety of quantitative MRI applications.

Automated summary

This study introduces a new approach for optimizing the data acquisition parameters of MR Fingerprinting experiments, utilizing B-splines to capture the structured relationships between parameters, resulting in improved computational efficiency. Through numerical simulations and experiments, it has been shown that the proposed method is effective and capable of solving the experimental design problem in a short amount of time.