Four Dimensional Magnetic Resonance Imaging With 3D Radial Sampling and Self-gating Based K-space Sorting: Early Clinical Experience on Pancreatic Cancer Patients

Abstract

Dynamic magnetic resonance imaging (MRI) has been used to characterize internal organ motion, but real-time acquisition has been limited to 2 dimensions (2D). Methods have been developed to reconstruct four dimensional MRI (4D-MRI) based on rebinning time-stamped 2D images or 2D K-space data. These methods suffer from anisotropic resolution and rebinning artifacts. We developed a 4D-MRI technique using 3D radial sampling and self-gating based K-space sorting (SG-KS-4D-MRI) to overcome these limitations, and applied it to study the target motion of pancreatic cancer patients. Ten patients were imaged during free breathing using 4D-CT, cine 2D-MRI and the SG-KS-4D-MRI method, which is an 8-minute spoiled gradient recalled echo (GRE) sequence with 3D radial-sampling K-space projections and 1D projection based self-gating. Tumor volumes were drawn at the end of exhalation phases in the SG-KS-4D-MRI and 4D-CT for each patient, and then mapped to the other phases using deformable registration with B-spline algorithm. Tumor volumes and motion trajectories were compared. An isotropic resolution of 1.6 mm was achieved in the SG-KS-4D-MRI images, which showed superior soft tissue contrast to 4D-CT and were free of visible rebinning artifacts. The tumor motion trajectory cross-correlations between SG-KS-4D-MRI and cine 2D-MRI directions in superior-inferior (SI), anterior-posterior (AP) and medial-lateral (ML) directions were 0.93 ± 0.03, 0.83 ± 0.10, and 0.74 ± 0.18, respectively. The tumor motion trajectory cross-correlations between the SG-KS-4D-MRI and 4D-CT in SI, AP and ML were 0.91 ± 0.06, 0.72 ± 0.18, and 0.44 ± 0.24, respectively. The average motion amplitude differences (Ddiff) between SG-KS-4D-MRI and cine 2D-MRI in SI, AP, and ML directions were 0.83 ± 0.52, 0.40 ± 0.17, and 0.48 ± 0.24 mm. The Ddiff between SG-KS-4D-MRI and 4D-CT in SI, AP, and ML directions were 1.09 ± 0.43, 0.53 ± 0.24, and 0.47 ± 0.26 mm, respectively. The average standard deviations of the GTV volumes (GTV_σ) calculated from each of the ten breathing phases were 0.8 cm3 (SG-KS-4D-MRI) and 1.0 cm3 (4D-CT), P = .004, indicating a statistically significant difference between the two imaging modalities. A novel 4D-MRI imaging sequence, SG-KS-4D-MRI, was used to evaluate target motion for free breathing pancreatic cancer patients. Pancreatic motion trajectories agreed well with 4D-CT and cine 2D-MRI data in the SI and AP directions, but agreement was limited in the ML direction, where image pixel size and displacements are of similar size. Use of SG-KS-4D-MRI results in improved spatial resolution, voxel isotropy, and gives an average motion pattern over an extended period of time; SG-KS-4D-MRI also results in statistically more consistent tumor volume definition across the ten breathing phases.