Impact of motion correction on reproducibility and spatial variability of quantitative myocardial T2 mapping

Sébastien Roujol, Tamer Basha, Sebastian Weingärtner, Mehmet Akçakaya, Sophie Berg, Warren J. Manning, Reza NezafatJournal of Cardiovascular Magnetic Resonance 2015 · DOI

Abstract

The ARCTIC technique substantially reduces spatial mis-alignment among T2-weighted images and improves the reproducibility and spatial variability of in-vivo T2 mapping.

Introduction

Quantitative $T_2$ mapping is a critical tool in Cardiac Magnetic Resonance (CMR) imaging, offering sensitivity to myocardial edema and inflammation—essential markers for diagnosing myocarditis and acute myocardial infarction. However, the accuracy of these maps is often compromised by physiological motion. Generating $T_2$ maps requires multiple images acquired at different echo times, which must be perfectly aligned. Respiratory and cardiac motion, particularly in patients unable to perform long breath-holds, leads to spatial misalignment and subsequent "blurring" or errors in the final quantitative map.

Methods

This study evaluates ARCTIC (Adaptive Registration of varying Contrast-weighted images for improved Tissue Characterization), a novel motion correction framework. Unlike standard registration algorithms that may confuse $T_2$ decay-induced signal changes with motion, ARCTIC employs a two-step process:

  1. Global Alignment: An initial affine transformation to correct large-scale displacements.
  2. Local Non-Rigid Correction: An extended optical flow formulation that simultaneously estimates motion fields and per-pixel intensity variations.

The framework was tested across three acquisition strategies: Breath-hold (BH), Free-breathing (FB), and Free-breathing with respiratory navigator gating (FB+NAV). To ensure clinical feasibility, the algorithm was implemented on a GPU, achieving a reconstruction time of approximately 20 seconds.

Experimental Design

The evaluation involved two cohorts:

  • Healthy Subjects (N=12): Used to assess the reproducibility of $T_2$ measurements across repeated scans.
  • Patient Cohort (N=50): Clinical patients referred for CMR, used to evaluate subjective image quality and algorithm performance in a real-world setting.

Results

The ARCTIC technique demonstrated significant improvements in both visual quality and quantitative metrics.

Comparison of T2-weighted images and T2 maps across different acquisition strategies (BH, FB, FB+NAV) before and after motion correction. Red arrows indicate motion artifacts in uncorrected images.

Improved Alignment and Precision

Motion correction significantly reduced Myocardial Boundary Errors (MBE) and increased the Dice Similarity Coefficient (DSC) across all acquisition types. For FB+NAV data, the MBE decreased from 0.81 mm to 0.63 mm ($p < 0.001$), indicating superior alignment of the myocardial walls across the image series.

Scatter plot comparing Myocardial Boundary Error (MBE) in pixels for uncorrected versus motion-corrected images. Points below the identity line indicate improved alignment.

Enhanced Reproducibility and Reduced Variability

In healthy volunteers, the variability between repeated $T_2$ maps (using 20 echo times) was nearly halved, improving from 3.9 ms to 2.2 ms ($p = 0.002$). Furthermore, spatial variability (noise within the myocardium) was reduced from 12.8 ms to 10.3 ms for 4-echo sequences.

Clinical Validation

A blinded cardiologist's assessment of the patient cohort showed a significant preference for ARCTIC-corrected maps, which scored 3.69 compared to 3.43 for uncorrected maps on a 4-point scale.

Discussion and Conclusion

While ARCTIC effectively addresses in-plane motion, it does not correct for through-plane motion (where the heart moves out of the imaging slice). Consequently, the use of respiratory navigators remains beneficial. The study demonstrates that with robust motion correction, free-breathing $T_2$ mapping can achieve a level of precision and reproducibility comparable to breath-holding techniques. This makes quantitative CMR more accessible and reliable for patients with heart disease who struggle with breath-holding. Future work will explore extending this framework to 3D mapping and other quantitative techniques like $T_1$ and $T_1\rho$ mapping.

Summary updated

Abstract

The ARCTIC technique substantially reduces spatial mis-alignment among T2-weighted images and improves the reproducibility and spatial variability of in-vivo T2 mapping.
BibTeX
@article{roujol2015impact,
  author = {Sébastien Roujol and Tamer Basha and Sebastian Weingärtner and Mehmet Akçakaya and Sophie Berg and Warren J. Manning and Reza Nezafat},
  title = {{Impact of motion correction on reproducibility and spatial variability of quantitative myocardial T2 mapping}},
  journal = {Journal of Cardiovascular Magnetic Resonance},
  year = {2015},
  doi = {10.1186/s12968-015-0141-1},
}
Figure