OpenMASC: An Open-Source Pipeline for Cross-Trajectory Metal-Aware Sampling and Correction in Accelerated MRI

2026-09-26

AI Summary

This study addresses MRI artifacts induced by metallic implants and the limitations of existing accelerated reconstruction methods, which rely on artifact-free data and lack paired datasets. To overcome these challenges, this work proposes a trajectory-agnostic joint optimization framework for acquisition and reconstruction. Methodologically, it introduces a fully open-source pipeline for generating physics-simulated paired CT/MRI data and presents MA-VarNet equipped with a DC Rectifier to correct artifacts. Furthermore, a reinforcement learning agent is designed to actively select k-space readouts, enabling decoupled yet synergistic training of the sampling strategy and the unrolled reconstruction network. Experimental results demonstrate that the proposed method is compatible with both Cartesian and radial trajectories, significantly outperforming conventional and learning-based baselines at 4× and 8× acceleration factors while effectively enhancing image quality.

Abstract

Metal implants corrupt MRI measurements throughout $k$-space, yet existing accelerated MRI methods assume clean data and most metal artifact reduction approaches assume fully sampled acquisitions. No public dataset provides paired $k$-space and images with and without metal for the same anatomy, and no framework jointly addresses artifact-aware acquisition and reconstruction across sampling trajectories. We present OpenMASC, an open-source pipeline covering the full workflow from data generation to deployment. A physics-based data generation module converts public CT volumes into paired clean and metal-corrupted MRI data in both Cartesian and radial formats. MA-VarNet, an unrolled reconstruction network with a per-cascade DC Rectifier, corrects artifacts that data-consistency steps reintroduce from corrupted measurements. A reinforcement learning agent actively selects $k$-space readouts and co-trains with the reconstruction network through a decoupled three-stage procedure. The framework is trajectory-agnostic except for the data-consistency operator, supporting both Cartesian and radial acquisition without architectural changes. Experiments on two datasets at $4\times$ and $8\times$ acceleration demonstrate consistent improvements over conventional and learned baselines on both trajectories.

Problem

• Metal artifact reduction
• Accelerated MRI
• k-space sampling
• Open dataset
• Cross-trajectory reconstruction

Innovation

• Metal Artifact Reduction
• Accelerated MRI
• Unrolled Reconstruction Network
• Reinforcement Learning
• Cross-Trajectory Sampling