Optimize Flip Angle Schedules In MR Fingerprinting Using Reinforcement Learning

2025-11-25

AI Summary

To address the challenges of complex flip-angle (FA) sequence design, insufficient fingerprint distinguishability in parameter space, and low acquisition efficiency in magnetic resonance fingerprinting (MRF), this work proposes a reinforcement learning (RL)-based FA scheduling optimization framework. We formulate FA sequence design as a sequential decision-making problem, integrating MRF transient signal dynamics modeling with exploration of high-dimensional parameter spaces to automatically discover optimal non-periodic FA sequences. Experiments demonstrate that the learned sequences significantly enhance fingerprint separability in the T₁/T₂ parameter space and enable accelerated imaging by reducing the number of repetition time (TR) intervals—while preserving quantitative accuracy. To our knowledge, this is the first systematic application of RL to MRF pulse sequence optimization. The proposed framework establishes a new paradigm for high-dimensional, nonlinear, multi-objective MR sequence design, effectively balancing imaging speed, robustness, and quantitative fidelity.

Technology Category

• Machine Learning: Reinforcement Learning
• Search and Optimization: Learning to Search
• Intelligent Robots: Learning & Optimization for ROB

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Magnetic Resonance Fingerprinting (MRF) leverages transient-state signal dynamics generated by the tunable acquisition parameters, making the design of an optimal, robust sequence a complex, high-dimensional sequential decision problem, such as optimizing one of the key parameters, flip angle. Reinforcement learning (RL) offers a promising approach to automate parameter selection, to optimize pulse sequences that maximize the distinguishability of fingerprints across the parameter space. In this work, we introduce an RL framework for optimizing the flip-angle schedule in MRF and demonstrate a learned schedule exhibiting non-periodic patterns that enhances fingerprint separability. Additionally, an interesting observation is that the RL-optimized schedule may enable a reduction in the number of repetition time, potentially accelerate MRF acquisitions.

Problem

• Optimizing flip angle schedules in MR fingerprinting
• Enhancing fingerprint separability across parameter space
• Reducing repetition time to accelerate MRF acquisitions

Innovation

• Reinforcement learning optimizes MR fingerprinting flip angles
• Learned non-periodic patterns enhance fingerprint separability
• Reduces repetition time to accelerate MRF acquisitions