INTRODUCTION

ECG is a widely used technique for cardiac motion gating in cardiac MR Imaging. However, the ECG signal is often impaired by the magnetohydrodynamic effect, particularly in high-field ($$$\geq$$$3T) scanners, for patients with hairy chests, and obese patients. As an alternative, self-gating provides a wireless method to obtain the cardiac dynamic state in 2D cine with about 8ms uncertainty under breath-hold¹. Nevertheless, compared to the standard ECG signal, the uncertainty of current 3D cardiac self-gating is approximately 20ms-30ms^2-4. This means that if a 50ms duration is set for a cardiac phase, the cardiac phase has an uncertainty of about half the cardiac phase duration.

METHODS

ROCK Sequence and in Vivo Experiments:

The 3D ungated spoiled gradient echo pulse sequence with a Rotating Cartesian K-Space (ROCK⁵) pattern is used in our study as shown in Fig. 1.

Experiments were conducted on 1 human (35-year-male) without contrast agent and 13 pigs (4-month-old) with gadolinium.
The sequence parameters for the human were TR/TE = 4.26/1.82 ms, RF = 5°, 400 Hz/px receiver bandwidth, matrix size of 256×256×80 mm³, 424 samples per readout, and spatial resolution of 1.25×1.25×1.5 mm³. The scan time was 8 minutes and 57 seconds.
The sequence parameters for the pigs receiving gadolinium were TR/TE = 3.52/1.4 ms, RF = 20°and 650 Hz/px receiver bandwidth, matrix size of 225×234×80 mm³, 370 samples per readout, and spatial resolution of 1.2×1.2×1.2 mm³. The scan time was 4 minutes and 26 seconds. Millimeter Wave Radar (mmWR⁶) is a non-contact respiratory triggering device and was recorded simultaneously with ECG signals.
All scans were performed on a 3T scanner (uMR890; United Imaging Healthcare, Shanghai, China) using a 12-channel chest coil array and a 16/48-channel spine coil array for signal reception.

Selective PCA self-gating method:

The 3D self-gating is usually implemented by interleaved k-readout lines through the k-space center along superior-inferior (SI) direction. The spatial variations in coil sensitivities and the heart’s motion make the SI projections obtained from different coils and locations have different weightings of cardiac and respiratory signals. To estimate the weightings of these two physiological signals, we introduced a parameter, Respiration to Cardiac Ratio (RCR) in the following equation
$$\mathrm{RCR}(n_c,n_x) = \frac{\int_{R_l}^{R_u}s(n_c,n_x,f)s_r(f)\mathrm{d}f}{\int_{C_l}^{C_u}s(n_c,n_x,f)s_c(f)\mathrm{d}f}.$$
Here, $$$s(n_c,n_x,f)$$$ represents the SI projection signal from $$$n_c$$$ channel and $$$n_x$$$ point at the frequency $$$f$$$; $$$R_l$$$ and $$$R_u$$$ denote the upper and lower cut-off frequencies for respiration and were set to 0.01Hz and 0.5Hz; $$$C_l$$$ and $$$C_u$$$ for cardiac signal were set to 0.5Hz and 2Hz for human and 0.5H and 3Hz for pigs. $$$s_r(f)$$$ and $$$s_c(f)$$$ are the estimated respiratory and cardiac spectrum.

Points ranking in the bottom 0.5% based on RCR were selected and normalized to generate the cardiac signal using PCA, depicted in Fig. 2 (red). Similarly, points ranking in the top 10% based on RCR underwent the same process to generate the respiratory signal, as shown in Fig. 2 (blue). RCR was calculated twice using updating $$$s_r(f)$$$ and $$$s_c(f)$$$. In the first calculation, $$$s_r(f)$$$ and $$$s_c(f)$$$ were generated from the mass center of SI projection, while in the second calculation, they were obtained from the PCA of selected SI points based on first calculated RCR.

RESULTS

The respiratory and cardiac self-gating from 1 human and 3 pigs is shown in Fig. 3. The averaged time delays (TDs) between ECG R-peaks and self-gating peaks were calculated and listed in Fig. 4. The selective PCA temporal precision achieved 11.43ms in total. The MR cine images of #pig 3 are shown in Fig. 5. Blood pool and cardiac muscle can be clearly seen in the cardiac phases.

DISCUSSION

The proposed RCR successfully selects respiration and heartbeat-related signals from mixed physiological signals in SI projections, accounting for coil sensitivity and heart location differences. This selection process enhances the cardiac signal while attenuating the respiratory signal, and vice versa for the respiratory signal. Compared with normal PCA self-gating methods, our method does not require a step to differentiate cardiac from respiratory gating signals. Since only 10% or 5% of the SI projection data are used for PCA, the calculation burden is highly reduced. During the experiment, the ECG had an R-peak detection failure rate of 35%, which strengthens the necessity for self-gating techniques.

CONCLUSION

Selective PCA offers reliable self-gating for 3D cardiac MR Imaging, accurate to 11ms, and applicable in challenging ECG-gating situations, such as hairy chests and obesity, as well as high-fields ($$$\geq$$$3T). The method not only pinpoints the gating moment but preserves the entire cardiac dynamics, paving the way for enhanced arrhythmia CMR Imaging.

Acknowledgements

No acknowledgement found.