INTRODUCTION

The reproducibility of scientific reports is crucial to advancing human knowledge. This abstract is a response to the 2023 ISMRM Challenge “Repeat it With Me: Reproducibility Team Challenge”. We reproduce the bSTAR sequence (1,2), and provide an open-source implementation.

bSTAR is a 3D half-radial dual-echo balanced steady-state free precession (bSSFP) sequence that has been proposed for free-breathing non-ECG-triggered thoracic imaging with extremely short TR at 1.5T (1,2). It is especially attractive for lung parenchyma imaging at low field due to the prolonged T2’ (3). bSTAR is suitable for various applications at low field strengths because a bSSFP sequence is instrumental in achieving high SNR compensating for reduced equilibrium polarization; and achieving a short TR for bSSFP is extremely important to avoid banding artifacts caused by enhanced concomitant fields.

We reproduce the bSTAR sequence for thoracic imaging at 0.55T using vendor neutral open-source frameworks to enable code sharing across different institutions, vendors, and scanner software versions. We used the Pulseq framework (4) for pulse sequence implementation, and Berkeley Advanced Reconstruction Toolbox (BART) for image reconstruction (5).

METHODS

Pulse sequence:
A free-running non-ECG-triggered bSTAR sequence was implemented with the Pulseq framework (4). A detailed pulse sequence diagram for bSTAR imaging is illustrated in Figure 1. Both wobbling Archimedean spiral pole trajectory (WASP) (3) and a spiral phyllotaxis trajectory (SP) (6) were implemented.

Pulseq implementation:
The Pulseq framework (abbreviated as Pulseq) defines RF, gradient, ADC, and delay events as basic components of a pulse sequence. Pulseq provides a Cartesian coordinate system [x, y, z] and referred to as Pulseq logical x y and z axes. We interpreted this coordinate system as vendor’s logical axes [RO, PE, SL] using “old/compat” option in the “Orientation mapping” parameter. Gradients and rotation matrices were defined using [PE, RO, SL] as the first, second, and third coordinate in a logical coordinate system to comply with our vendor’s coordinate transformation from a logical coordinate system to a physical coordinate system. A gradient event on different axis is considered as a separate event.

Experiments:
All imaging experiments were performed on a whole-body 0.55T scanner (prototype MAGNETOM Aera; Siemens Healthineers, Erlangen, Germany) with gradients capable of 45mT/m amplitude and 200 T/m/s slew rate. A six-element body coil (anterior) and six elements from an 18-element spine coil (posterior) were used for signal reception. The default shim setting (tune-up mode) was used. Two healthy volunteers (1 male and 1 female) were scanned under a protocol approved by our institutional review board after providing written informed consent.

Phantom study:
An ISMRM/NIST system phantom (7) was scanned with WASP and SP trajectories. Imaging parameters were: TR = 1.38 ms, TE1 = 0.13 ms, TE2 = 1.17 ms, RF duration = 200 µs, FA = 25°, FOV = 340 x 340 x 340 mm³, isotropic resolution = 1.6 x 1.6 x 1.6 mm³, bandwidth = 1929 [Hz/px], number of interleaves = 89/88, half-radial projections = 31152, and total scan time = 43 sec.

Human study:
In one subject, a 50-sec breathhold acquisition during end-expiration was performed using the SP trajectory with 89 interleaves (39961 half-radial projections). This volunteer was capable of very long breatholds in order to facilitate retrospective undersampling experiments that are planned. In one subject, a 23-sec breathhold acquisition during end-expiration was performed using the WASP trajectory with 4 interleaves (17000 half-radial projections).

Trajectory measurements:
K-space trajectories along +X, -X, +Y, -Y, +Z and -Z physical axes were measured with Duyn’s method (8) and a recently proposed method by Zhao et al. (9). 3D radial half-spokes were calculated with a linear combination of measured trajectories.

Reconstruction:
Image reconstruction was performed with compressed sensing SENSE reconstruction implemented in the Berkeley Advanced Reconstruction Toolbox (BART). A wavelet transform (Daubechies 2) was used for a sparse transform with a regularization parameter of 0.005.

RESULTS

Figure 2 reproduces Figure 3 of Ref. 2 using an ISMRM/NIST system phantom. Image reconstructions from both echoes show no visible artifacts including banding artifacts and geometric distortion due to inaccuracies in k-space trajectories.

Figure 3 reproduces Figure 4 of Ref. 1 using an ISMRM/NIST system phantom. WASP trajectories with 88 and 89 interleaves did not create noticeable eddy current artifacts, demonstrating its flexibility in the design of 3D radial trajectory patterns.

Figure 4 shows the exemplary bSTAR images of a male volunteer acquired during end-expiratory brearthhold. Banding artifacts are not visible within the FOV of interest.

Figure 5 compares image reconstructions by Pulseq bSTAR and original bSTAR (implemented in Siemens’ IDEA programming language). Each method acquired data separately with its own pulse sequence and reconstructed images with its reconstruction pipeline. DICOM images were created at the end of each reconstruction pipeline and compared in a DICOM reader.

DISCUSSION & CONCLUSION

We have successfully reproduced the bSTAR method, and figures with comparable image quality compared to published literature (1,2). This study also demonstrates the power of open-source frameworks, specifically Pulseq, because designing a pulse sequence in a vendor proprietary environment requires expertise and tremendous effort.