Dual Echo TIme Cone (ETICone): A 3D-Cone Trajectory for Sampling-Efficient Free-Running bSSFP Cardiac MRI at 0.55 T

Background: Balanced SSFP (bSSFP) is highly attractive at low field because of favorable T2/T1 contrast and reduced banding sensitivity, which allows prolonged TR. Free-running (FRF) 3D bSSFP cardiac MRI is often performed with radial trajectories due to their robustness to motion and efficient low-frequency k-space sampling [1]. However, radial oversamples central k-space while sparsely covering high-frequency regions, resulting in a degraded point spread function (PSF) and visible streaking in reconstructed images. To address these limitations, we developed the Dual Echo TIme Cone (ETICone) trajectory, which improves high-frequency coverage and PSF homogeneity while incorporating bipolar gradient moment nulling to reduce flow sensitivity.

Methods:
The ETICone readout consists of outward and symmetric inward branches connected with cosine-bell weighting (Fig. 1). Readouts were rotated to cover k-space using WASP [2], with additional randomized rotations around each cone axis (Fig. 1). The cone aperture and the number of revolutions were optimized by minimizing the peak-to-sidelobe ratio of the PSF. Time-optimized gradient waveforms were designed using the minTimeGrad algorithm [3] with system constraints.
Experiments were performed on a whole-body 0.55 T MR system using both custom radial and ETICone FRF bSSFP sequences with parameters described in Table 1. ETICone trajectory was implemented in Pulseq [4]. Images were reconstructed using 5D Compressed Sensing (CS) [1]. Self-gating supported respiratory binning (4 phases) and cardiac binning (12 phases). Image sharpness was quantified at end-diastole using the Tenengrad focus measure [5] on a central cubic ROI, with statistical comparisons across slices using Wilcoxon signed-rank tests. Images were evaluated for CR between left ventricle blood pool and myocardium and for SNR on left ventricle blood pool along cardiac dimension.





Results: ETICone exhibited a more homogeneous PSF than radial, leading to visibly reduced streaking in reconstructed images (Fig. 2). At end-diastole, sharpness was significantly higher with ETICone compared to radial (986±176 vs.728±112; p< 0.001). Across cardiac phases, ETICone also provided higher SNR (16.2±1.5 vs. 9.9±1.6) and similar contrast ratio (CR) between blood and myocardium (0.426±0.077 vs. 0.411±0.147; p=0.58). Bipolar gradient re-winders enabled simultaneous nulling of zeroth and first gradient moments, which mitigated but did not fully remove flow-induced signal loss.

Conclusion: The proposed ETICone trajectory enables efficient FRF 3D imaging at low field, with improved high-frequency k-space sampling, reduced streaking, and significantly improved sharpness and SNR compared to radial acquisitions, while maintaining a similar blood-to-myocardium CR. A limitation of this study is that CS regularization weights were optimized for the radial trajectory and applied identically to ETICone. The dual-echo capability also provides opportunities for fat–water separation.

Acquisition parameters for Radial and ETICone free-running bSSFP sequences.

	Radial	ETICone
TR (ms)	5.25	4.75
TE1/TE2 (ms)	2.63	0.29/4.75
Isotropic res (mm)	1.5
Tot. Acq. (min:sec)	6:00
FOV (mm)	300
Dwell time (s)	10⁻⁵
N readouts(10³)	71	84
Flip Angle (deg)	65

ETICone (top) and Radial (bottom) sequence block diagram (left) and corresponding k-space trajectory (right).
Comparison of the proposed ETICone trajectory (left) and conventional Radial trajectory (right) for free-running bSSFP at 0.55 T. Top: corresponding point spread functions (PSF), log-scaled. Bottom: end-diastole 5D reconstructions with total variation regularization, shown in axial and coronal views. Thanks to ETICone improved $k$-space sampling over radial, undersampling PSF is more homogeneous, leading to reduced streaking, and improved image sharpness.