Gradient system characterization of a 1.5 T MR-Linac with application to 4D UTE imaging for adaptive MR-guided radiotherapy of lung cancer

Purpose

To measure the gradient system transfer function (GSTF) of an MR-Linac (Elekta Unity, Stockholm, Sweden) using an accessible phantom-based method and to apply trajectory corrections for UTE image reconstruction in the context of MR-guided radiotherapy of lung cancer.

Methods

The first-order GSTF of a 1.5 T, split gradient Elekta Unity MR-Linac was measured using a thin-slice technique to characterize gradient system imperfections for each physical gradient axis (X, Y, Z). Repeatability measurements of the GSTF were performed 48 h apart. The GSTF was applied to trajectory correction in multi-echo UTE image reconstruction (TEs = 0.176, 1.85, 3.52 ms) to allow for UTE-Dixon inputs in the generation of synthetic CT. Images were acquired in an anthropomorphic phantom and in two free-breathing lung cancer patients. For patient scans, respiratory-correlated 4D-MR images were reconstructed using self-navigation and an iterative compressed-sensing algorithm.

Results

The GSTF magnitude was similar across the X/Y/Z axes up to ˜6 kHz. The GSTF phase was similar between the X/Y/Z components up to ˜3 kHz. Repeatability measurements demonstrated minimal variations corresponding to a system delay difference of 0.06 μs. Corrected UTE trajectory spokes are shifted approximately 1 m⁻¹ compared to the nominal k-space location. Corrected phantom and patient UTE images exhibited improved signal uniformity and contrast and reduced halo and signal loss artifacts. Trajectory correction for the later TE images did not improve overall image quality.

Conclusion

The proposed GSTF measurement method using standard MR-Linac hardware enables successful trajectory correction in UTE imaging reconstruction, with applications to lung synthetic CT generation for MR-guided radiotherapy.