Cylinder 3D radial acquisition for reduced imaging artifacts and better resolution at 1.5 T
Yajing Zhang1, Jiazheng Wang2, and Chenguang Zhao1

1Philips Healthcare, Suzhou, China, 2Philips Healthcare Greater China, Beijing, China


We have developed a novel 3D radial sequence for motion insensitive MRI, which replaces the frequency encodings in the radial plane in prior-art stack-of-star sequences with stepwise phase encodings to reduce the streaking artifacts that can arise from chemical shifts and system imperfections. The sequence achieved better image homogeneity with less imaging artifacts when compared to the prior-art sequence at 1 mm isotropic resolution with golden angle acquisition, both in phantom and in human brain and abdomen imaging.


Stack-of-star (SoS) radial imaging has been established as a standard technique for T1 weighted imaging in head and abdomen1 for its motion immunity when Golden Angle scheme is used together with its spoke acquisition2. However, as the readout direction rotates, the k-space trajectory and phase errors, arising from gradient chain delays, eddy currents (EC), chemical shifts and B0 field inhomogeneity, would lead to misaligned radial k-space, and consequently image inhomogeneity and streaking artifacts after gridding recon. These artifacts are particularly common in imaging areas such as abdomen and cardiovascular, despite the pre-reconstruction efforts on the clinical scanners to remove the 0th, 1st and high orders phase errors. Here we present a cylinder-like acquisition trajectory such that each 2D radial k-space plane is stepwise phase-encoded, with infinite bandwidth and hence improved immunity to system imperfections. A similar method has been used in a volumetric scanning for motion detection and compensation3. In the present study, we focus on exploring the advantages of cylinder scan in comparison to SoS. Experiments were performed on phantom and human subjects. Results indicated that the cylinder methods was helpful in reducing streaking artifact level and improving image uniformity.


The prior-art SoS 3D k-space and the proposed alternative are depicted in Figure 1A and 1B, respectively. In prior-art (Figure 1A), each kxy plane contains multiple readouts, where each readout passes the 2D k-space center and is rotated by a golden angle from its adjacent readout, while the 3rd dimension is filled with stepwise phase encoding. In the proposed cylinder trajectory (Figure 1B), the readout direction coincides with the 3rd dimension while the 2D radial plane is stepwise phase-encoded, and each two adjacent Cartesian plane is still rotated by a golden angle. The proposed trajectory was implemented into a spoiled gradient echo sequence on a 1.5 T Multiva system (Philips Healthcare Suzhou, China) to compare with the prior art trajectory, and experiments were performed on both phantom and healthy volunteers with informed consent (on both head and abdomen). The imaging protocol was approved by a local Ethical Review Body, with 4.2 ms TR, 1.7 ms TE, and 110% angle percentage for both trajectories to achieve 1 mm isotropic voxel size. The angle percentage describes the ratio between the number of angles in the radial plane and the required number of phase encoding lines for the same plane if acquired with Cartesian k-space. Golden angle was used such that each two adjacent Cartesian planes were separated by 111.246°. Data acquired from both methods went through the gridding reconstruction and channel combination implemented in Philips recon 2.0 (Philips Healthcare Best, the Netherlands).


Exemplary phantom images from the SoS and the cylinder acquisitions are shown in Figure 2A and 2B, respectively. SoS images turned out to have apparent signal inhomogeneity, streaking artifacts (short arrow), and blurring of fine structures (long arrow), which are much less observable in the cylinder images. Figure 2C and 2D demonstrate the k-space lines acquired from both methods that were used to fill the radial plane in the middle of the 3rd dimension. Each column in the figures is a spoke in the radial plane, with phase being color-coded, showing better alignment of k-space lines in the proposed trajectory. Figure 3A, 3B, and 3C show the brain images from the SoS without phase correction, the cylinder acquisition, and the SoS with phase correction, respectively. While phase correction improved image uniformity for SoS acquisition, the loss of fine structures sustained in areas like frontal sinus (arrow). These problems were not observed in the cylinder image (Figure 3B). Figure 3D and 3E display the images acquired on human abdomen with free breathing, acquired with SoS (phase corrected) and cylinder respectively. While both methods showed motion insensitivity, the cylinder image was less contaminated by streaking artifacts.

Discussion and Conclusion

The proposed cylinder trajectory reduces streaking artifacts and improves true image resolution when compared to the prior-art stack-of-star trajectory, while the motion immunity and the imaging speed was retained. This technique holds clinical potential particularly for regions where motion insensitivity is preferred and local B0 field inhomogeneity presents. Development of trajectory-specific phase correction method would be favorable for the proposed acquisition scheme in the future to further improve imaging quality.


No acknowledgement found.


1. H. Chandarana, et al. Free-breathing radial 3D fat-suppressed T1-weighted gradient echo sequence: a viable alternative for contrast-enhanced liver imaging in patients unable to suspend respiration. Invest. Radiol. 2011; 46: 648-653.

2. S. Winkelmann, et al. An Optimal Radial Profile Order Based on the Golden Ratio for Time-Resolved MRI. IEEE Trans Medical Imaging 2007; 26 (1)

3. W. Lin. et al. Fast, Robust and Self-Navigated 3D Cylindrical Imaging: MP-RAGE and FLAIR. ISMRM.


Figure 1. Cylinder and SoS acquisition methods with readout (M), phase encoding (P) and slice (S) directions denoted. Left: Stack-of-star acquisition, rotation around slice direction. Right: Cylinder acquisition. Rotation around readout direction.

Figure 2. Comparison between Radial and Cylinder acquisitions. a. Radial phantom image. b. Cylinder phantom image. c. k-space lines of Radial. d. k-space lines of cylinder.

Figure 3. Comparison between Radial and Cylinder scans in Head and Abdomen. a. Radial head without phase correction. b. cylinder head imaging. c. Radial head imaging with phase correction. d. radial abdomen imaging. e. cylinder abdomen imaging.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)