Design and application of an MR reference phantom for multicentre lung imaging trials

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Radiology

Background:

• Multicenter lung imaging studies using MRI are increasing.
• A need exists for standardized phantoms to assess and compare MRI image quality across different vendors and institutions.
• Commercially available phantoms for this purpose are currently lacking.

Purpose of the Study:

• To design and implement a cost-effective and user-friendly reference phantom for MRI lung imaging.
• To address the specific requirements of imaging the lungs with MRI in multicenter settings.

Main Methods:

• A phantom was designed to simulate four compartments: lung, blood, muscle, and fat, relevant to proton-MRI of the chest.
• Multiple phantom instances were measured at 15 sites using a proton-MRI protocol for an in vivo COPD study.
• Image quality metrics including signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), structure, and edge depiction were extracted.

Main Results:

• Low intra-scanner variability was observed for SNR (4.5% in lung, 4.0% blood, 3.3% muscle, 3.7% fat).
• Substantially higher inter-scanner variability was found for SNR (41% lung, 32% blood, 27% muscle, 32% fat).
• Significant variations in structure and edge depiction were noted among different scanner types and even among identical models with different gradient systems.

Conclusions:

• The developed reference phantom reliably quantified image quality aspects and identified significant inter-scanner variability in pulmonary MRI studies.
• Variability was more pronounced in lung tissue compared to other simulated tissues.
• Reference phantoms are essential for detecting bias in multicenter in vivo studies and harmonizing equipment or data.