Free breathing myocardial perfusion MRI with SW-CG-HYPR using motion correction

  • Lan Ge
  • Aya Kino
  • Mark Griswold
  • James Carr
  • Debiao Li
Open Access
Poster presentation
  • 824 Downloads

Keywords

Motion Correction Breath Hold Free Breathing Image Quality Score Reduce Motion Artifact 

Introduction

The diagnostic value of first-pass perfusion MRI is limited by the low spatial coverage, resolution, SNR, and motion artifacts. Sliding-Window Conjugate-Gradient HYPR [1] (SW-CG-HYPR) has been proposed to acquire perfusion images with increased spatial coverage, better spatial resolution, and improved SNR [2]. However, this method is sensitive to the respiratory motion and thus limited to the breath hold. Motion correction may be useful to reduce motion artifacts and allow for free-breathing first-pass perfusion.

Purpose

To develop and test a non-model-based motion correction method combined with SW-CG-HYPR to perform free-breathing myocardial MR imaging.

Methods

An ECG-triggered, multi-slice FLASH sequence with radial sampling was used. As shown in Figure 1, radial sampling was applied in a segmented interleaved fashion. Multiple slices were acquired after each saturation recovery pre-pulse. The motion correction method is illustrated in Figure 2. Both translation and rotation of the heart were detected in image domain by calculating the normalized cross-correlation coefficients. Motion correction was performed in k-space by rotating the undersampled k-space and shifting the phase by a factor of exp(-2πi(δx/Nready/Npe)), where δx and δy are the number of pixels to shift in × and y direction, and Nread and Npe are the total number of pixels along readout and phase encoding direction. Sliding window was used to reconstruct the composite images, and the time-resolved images were reconstructed after CG-HYPR processing. Six healthy volunteers were scanned using a 1.5 T system, with and without breath hold, during first-pass of the contrast agent. Imaging parameters included: TR/TE/flip-angle = 3.2/1.6 ms/10º, spatial resolution = 1.3 × 1.3 × 10 mm3, and number of slices = 6. The images were qualitatively graded by a reviewer using a score of 1-4 (1: worse; 4: best), and the signal changes vs. time curves were compared.
Figure 1

Schematic of the myocardial perfusion acquisitions sequence. Radial k-space is highly undersampled, interleaved and equally space. Mulitple slices are acquired after each SR preparation pulse. In this work, two slices were acquired after each SR pre-pulse, and totally six slices were acquired for each cardiac cycle.

Figure 2

Schematic diagram illustration the motion correction scheme. The ROI (illustrated by the red rectangular) of the low resolution imates (b), reconstructed from the center of the undersampled k-space, is compared with the composite image (without motion correction) (a) for the motion detection. After the motion correction the k-space is the precessed by SW-CG-HYPR method for the time-resolved myocardial perfusion images.

Results

The average image quality score of the free-breathing images with motion correction (3.09 ± 0.37) is significantly higher than those without motion correction (2.26 ± 0.40), and is comparable to the successful breath-holding images (3.10 ± 0.41) (Figure 3). The signal changes in motion corrected free-breathing images were closely related to those in the breath-holding images, with a correlation coefficient of 0.9764 for myocardial signals (Figure 4 and 5).
Figure 3

Comparision example (a) and mean image quality scores (b) of free-breathing images without motion correction, free-breathing images with motion correction and breath-holding images.

Figure 4

One example comparion of left ventericular and myocardial signal intensity changes between free-breathing images with motion correction and breath-holding images during first-pass perfusion. A close correlation between the two datasets is observed.

Figure 5

Correlation between signal intensities of free-breathing images with motion correction and those breath-holding images for all of the volunteers. The two datasets are highly correlated, with a correlation coefficient of 0.9764

Conclusion

The image quality of myocardial perfusion MRI using SW-CG-HYPR was substantially improved after motion correction. This technique may allow myocardial perfusion MRI during free breathing.

References

  1. 1.
    Griswold MA, et al: #834. Proceedings of ISMRM. 2007Google Scholar
  2. 2.
    Ge L, Li D, et al: Magn Reson Med. 2009, DOI: 10.1002/mrm.22059Google Scholar

Copyright information

© Ge et al; licensee BioMed Central Ltd. 2010

This article is published under license to BioMed Central Ltd.

Authors and Affiliations

  • Lan Ge
    • 1
  • Aya Kino
    • 1
  • Mark Griswold
    • 2
  • James Carr
    • 1
  • Debiao Li
    • 1
  1. 1.Northwestern UniversityChicagoUSA
  2. 2.Case Western Reserve UniversityClevelandUSA

Personalised recommendations