Cardiac quantitative susceptibility mapping (QSM) for heart chamber oxygenation

Abstract

Purpose: To demonstrate the feasibility of in vivo quantitative susceptibility mapping (QSM) in cardiac MRI and to show that mixed-venous oxygen saturation (SvO₂ ) can be measured non-invasively using QSM.

Methods: Electrocardiographic-gated multi-echo 2D gradient echo data were collected at 1.5 T from 14 healthy volunteers during successive breath-holds. Phase wraps and fat chemical shift were removed using a graph-cut-based phase analysis and IDEAL in an iterative approach. The large susceptibility range from air in the lungs to blood in the heart was addressed by using the preconditioning approach in the dipole field inversion. SvO₂ was calculated based on the difference in blood susceptibility between the right ventricle (RV) and left ventricle (LV). Cardiac QSM quality was assessed by two independent readers.

Results: Nine out of fourteen volunteers (64%) yielded interpretable cardiac QSM. QSM maps showed strong differential contrast between RV and LV blood with RV blood having higher susceptibility values (291.5 ± 32.4 ppb), which correspond to 78.3 ± 2.3% SvO₂ .

Conclusion: In vivo cardiac QSM is feasible and can be used to measure SvO₂ , but improvements in data acquisition are needed. Magn Reson Med 79:1545-1552, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Keywords: cardiac magnetic resonance imaging (cardiac MRI); mixed venous oxygen saturation (SvO2); quantitative susceptibility mapping (QSM).

Figure 1

Flow chart of cardiac QSM post-processing.

Figure 2

Two representative high quality cardiac QSM examples, demonstrating differential susceptibility between the left and right sides of the heart. Magnitude images reformatted into long axis in the top row, short axis magnitude images in the middle row, and corresponding QSM in the bottom row. QSM demonstrated oxygenation differences throughout the basal (left), mid, and apical (right) chambers, in both examples.

Figure 3

A representative cardiac QSM example with reasonable quality. Differential susceptibility between the left and right sides of the heart can be seen. Magnitude images reformatted into long axis in the top row, short axis magnitude images in the middle row, and corresponding QSM in the bottom row. Columns represent basal (left), mid, and apical (right) chambers.

Figure 4

Representative cardiac QSM example with poor quality. The susceptibility contrast is mostly degraded by the shadowing artifacts. The long axis magnitude image (top) showed that there were a few misregistered slices (red arrows), suggesting inconsistent breath holding positions.

Figure 5

Comparison of R2* map, and QSM maps computed using PDF+MEDI and TFI algorithms. In this example, both QSM maps demonstrated much improved blood contrast between the right and left ventricles than the R2* map does. Between the two QSM maps, the result from TFI has better RV to LV contrast. The susceptibility difference between RV and LV blood was 327.4 ppb (corresponding to 77.0% SvO2) with TFI, and 138.6 ppb (90.1% SvO2, well above the normal range) with PDF+MEDI.