INTRODUCTION

Effective radiation therapy relies on high geometric integrity of the MR image information used for treatment planning. Apart from gradient system non-idealities, main magnetic field inhomogeneities and patient-induced B0 susceptibility artefacts potentially lead to geometric distortion¹. B0 susceptibility is field strength dependent², and lower field strengths result in overall reduced absolute susceptibility³. The goal of this study is to analyze the effect of lower susceptibility on geometric image integrity of a conventional GRE and an echo-planar diffusion-weighted sequence by means of both digital simulations and in vivo measurements.

METHODS

Simulations: Bloch simulations of a 2D GRE sequence (voxel size = 1.5x1.5x3mm³, FoV = 288x288mm³, acquisition matrix = 192x192, bandwidth = 350Hz/px) and a digital phantom⁴ were performed at 1.5T (TE = 3.5ms, TR = 7ms, flip angle = 20°) and 0.55T (TE = 4ms, TR = 8.9ms, flip angle = 30°). At both field strengths, two types of simulations were performed. First, an ideal system (main magnetic field, RF field, gradient fields, RX profile) is assumed, and B0 susceptibility effects are turned off (termed “Simulation ideal”). The second simulation applied similar parameters, but main magnet field inhomogeneities as provided by the manufacturer and B0 susceptibility effects are turned on (termed “Simulation B0”). The effect of B0 shimming was not simulated.

In vivo Study: Three healthy volunteers (2 female, 1 male, age: 67±4 years, weight: 78±15 kg) were imaged at a 1.5T (MAGNETOM Sola, Siemens Healthcare, Erlangen, Germany) and a prototypical version of a 0.55T scanner (MAGNETOM Free.Max, Siemens Shenzhen Magnetic Resonance Ltd., Shenzhen, China). Measurements at both field strengths covering the optic nerve were performed directly one after another by means of dedicated head coils. A 3D T2 turbo-spin-echo sequence (SPACE) was used to acquire high-resolution data (voxel size = 1x1x1mm³, FoV = 268x268x176mm³, acquisition matrix = 256x256x176) at 1.5T (TE = 292ms, TR = 1900ms, echo spacing = 3.28ms, echo train duration = 758ms, acquisition time = 3min 13s) and 0.55T (TE = 412ms, TR = 3200ms, echo spacing = 4.68ms, echo train duration = 1076ms, acquisition time = 6min 38s). A high readout bandwidth of 651Hz/px was applied, to minimize image distortion due to B0 inhomogeneity. Echo-planar diffusion-weighted images were obtained at both field strengths, and Apparent Diffusion Coefficient (ADC) maps were calculated. Parameters that influence geometric distortions due to B0 inhomogeneities were matched (voxel size = 1.4x1.4x4mm³, acquisition matrix = 192x192x24, bandwidth = 766Hz/px, echo spacing = 1.53ms). Data acquisition in k_y direction was accelerated two-fold using GRAPPA. The remaining imaging parameters included FoV = 268x268x119mm³, TE = 119ms, TR = 5300ms, b-value 1 = 0 s/mm², b-value 2 = 1000 s/mm², EPI factor = 154, and acquisition time = 2min 56s at both field strength. Additionally, quantitative B0 off-resonance field maps were calculated based on the phase differences of a 2-echo GRE sequence.

Evaluation: Gradient non-linearities have the same impact on geometric integrity of the in vivo EPI and SPACE images. Therefore, non-rigid registration between ADC and T2 SPACE reference images was performed. Mean and standard deviation of the resulting distortion vectors across all volunteers were used as a measure of ADC map distortion due to B0 field inhomogeneities and susceptibility. Additionally, we report mean and standard deviation of the minimal, maximal, and interquartile range (IQR) B0 off-resonance values across all volunteers in Hz in a slice covering the optic nerve.

RESULTS AND DISCUSSION

Figure 1 depicts reconstructed images of the performed Bloch simulations for 1.5T (left) and 0.55T (right). Smaller overall distortion in the simulations that consider B0 field inhomogeneities (“Simulation B0”) compared to the ideal simulations can be observed at 0.55T compared to 1.5T.
Figure 2 depicts resulting images of the described in vivo study. Qualitatively, ADC maps acquired at 0.55T allowed better delineation of structures in high-susceptibility regions, such as the optic nerve. Quantitatively, imaging at 0.55T resulted in reduced mean ADC map distortion in a slice covering the optic nerve across all volunteers (1.8±0.3 mm) compared to ADC maps acquired at 1.5T (4.6±0.3 mm). Minimal, maximal, and IQR off-resonance B0 map values of -40±8 Hz, 56.3±9 Hz, and 13±2 Hz (at 1.5T) as well as -19±11 Hz, 28±4 Hz, and 5±1 Hz (at 0.55T) indicate smaller B0 map variance at low field compared to higher fields.

CONCLUSION

Imaging with less geometric distortion due to B0 susceptibility artefacts to support accurate MR in RT planning is achievable at 0.55T compared to higher field strengths.

Acknowledgements

No acknowledgement found.