Technical challenges of quantitative chest MRI data analysis in a large cohort pediatric study

Area of Science:

• Radiology
• Medical Imaging
• Quantitative MRI

Background:

• Geometric distortion (GD) in Magnetic Resonance Imaging (MRI) can impact the accuracy of lung volume quantification.
• Accurate lung segmentation is crucial for quantitative analysis in various thoracic imaging applications.
• Evaluating different segmentation methods is essential for optimizing MRI-based lung volume measurements.

Purpose of the Study:

• To assess the impact of geometric distortion on MRI-based lung volume quantification.
• To compare the performance of manual, semi-automated, and fully automated lung segmentation techniques.
• To determine the feasibility of using chest MRI for radiation-free lung volume quantification in large cohort studies.

Main Methods:

• Phantom studies were performed using MRI and CT to quantify GD, with CT serving as the gold standard.
• Dice scores were utilized to evaluate shape overlap between segmented lung volumes.
• Eleven subjects underwent multiple MRI acquisitions, and five segmentation methods were tested on 44 scans, employing statistical analyses including ICC and Bland-Altman plots.

Main Results:

• MRI overestimated lung volume compared to CT by 5.56-6.99%, varying with MRI position and Gradwarp correction.
• 3D Gradwarp MRI images demonstrated higher Dice scores and reduced intra-object differences.
• Semi-automated and fully automated segmentation methods showed high agreement with manual segmentation (ICC 0.971-0.995) and significantly reduced segmentation time.

Conclusions:

• Geometric distortion in MRI lung volume quantification is measurable and dependent on imaging parameters.
• Automated segmentation tools offer accurate, reproducible, and time-efficient lung volume quantification.
• Chest MRI represents a valuable radiation-free alternative to CT for quantitative lung analysis in large-scale research.