Adaptive Plane Reformatting for 4D Flow MRI using Deep Reinforcement Learning

2025-05-31

ISMRM Annual Meeting

AI Summary

Manual plane reformatting of 4D flow MRI is time-consuming, subjective, and existing deep reinforcement learning (DRL) methods suffer from poor generalization—exhibiting sharp performance degradation under anatomical position or orientation shifts. Method: We propose the first adaptive DRL framework for 4D flow MRI plane reformatting, integrating Proximal Policy Optimization (PPO), differentiable geometric coordinate transformations, and a multi-center, multi-vendor data-driven simulation training environment to enable anatomy-free, end-to-end slice localization. Contribution/Results: This work pioneers DRL application to this task and supports robust optimization under flexible coordinate systems. Experiments demonstrate sub-pixel accuracy in complex flow regions—including the aorta and pulmonary arteries—matching gold-standard landmark-based methods. The framework significantly improves cross-scanner generalizability and clinical deployability.

Technology Category

• Machine Learning: Reinforcement Learning
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Search and Optimization: Sampling/Simulation-based Search

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: On-Device user modeling, personalization, and recommendation

Abstract

Motivation: The standard approach for plane reformatting in 4D flow MRI is manual, leading to time-consuming and user-dependent results. Goal(s): Our goal was to enhance plane reformatting in 4D flow MRI and overcome limitations associated with existing automated methods. Approach: We introduce a novel approach that employs deep reinforcement learning (DRL) with a flexible coordinate system for precise and adaptable plane reformatting. Results: Results demonstrate superior performance compared to baseline DRL and similar outcomes compared to those of landmark-based techniques, showing its potential for use in complex medical imaging scenarios beyond 4D flow MRI. Impact: The proposed framework allows for automated, precise, and adaptive plane reformatting, facilitating the use of 4D flow MRI in clinical routines. It was trained with data sets from different vendors, making this approach widely applicable.

Problem

• Adaptive plane reformatting in 4D Flow MRI using DRL
• Overcoming position/orientation dependency in test datasets
• Improving accuracy in angular/distance errors for flow measurements

Innovation

• Uses deep reinforcement learning for plane reformatting
• Employs flexible coordinate system for any orientation
• Adopts A3C algorithm outperforming DQN