0931
Radial streak artifact reduction in multi-contrast imaging and parameter mapping using beamforming1Medical Imaging, University of Arizona, Tucson, AZ, United States, 2Electrical and Computer Engineering, University of Arizona, Tucson, AZ, United States
Synopsis
Streaking artifacts are a common source of image quality degradation in radial MRI even with sufficient sampling. While coil removal is popularly used to mitigate streaking artifacts, this method is known to suffer from undesirable signal loss. Recently, a streak artifact reduction method has been proposed that offers a better balance between artifact reduction and signal retention. This approach endows the channel combination maps with the ability to suppress streaking artifacts and is implemented as a post-processing step. In this work, we present a computationally efficient refinement of this approach and extend it to multi-contrast imaging and parameter mapping applications.
Introduction
Radial MRI is becoming increasingly popular but streaking artifacts arising from gradient nonlinearities can corrupt the entire image even with sufficient sampling1 and several approaches have been proposed to reduce them1-8. This problem is particularly acute in large FOV applications like abdominal MRI. One solution to this problem involves removing the subset of coils in the phased-array that are significantly corrupted by streaks before coil combination1,3. While automatic approaches to accomplish this exist, they can suffer from signal loss in some cases. Recently, a simple post-processing approach, B-STAR, to reduce streaking artifacts based on phased array beamforming9,10,11 (spatial filtering) was proposed8. This approach uses streak artifact information in the form of an interference correlation matrix and helps suppress streaks during coil combination.Beamforming for interference suppression
The B-STAR approach was motivated by the adaptive coil combination (ACC) scheme for noise suppression9. Both these approaches seek to optimize the signal to noise (interference) metric (SNR/SINR) in a spatially varying manner and require repeated eigenanalysis across the imaging scene. When coil sensitivity information is already available, a computationally efficient approach12 for B-STAR exists and is outlined in Figure 1(A). Note that B-STAR relies on the extraction of interference samples (streaks) from all the coils to generate an interference covariance matrix (Figure 1(B)). This matrix is diagonally loaded12 to allow a tradeoff between artifact suppression and signal retention (Figure 1(A)). The result of the B-STAR approach is a set of coil combination maps which act on the coil images to suppress streaking artifacts. An illustration of this artifact suppression is presented in Figure 1(C).
Figure 2(A) reiterates the B-STAR approach for streak artifact reduction. In higher dimensional imaging applications like dynamic imaging/parameter mapping, images sampled at different temporal points may exhibit different streaking behavior. While applying B-STAR for each time point in the temporal dimension is possible, the computational cost scales linearly with time points. Note however, that the suppression of artifact is tied to correlations across the coil dimension (and not space/space-time). An alternative approach is to form a space-time interference matrix by pooling all the interference samples from across space-time and creating a single set of channel combination maps (Figure 2(B)).
Methods and Results
Data from a radial fast spin echo (radFSE) pulse sequence were acquired using a 1.5T Siemens scanner on five subjects with echo spacing=7.3ms, 192 views with 256 readout points/view, echo train length=32, TR=4000ms, slice thickness=8mm, 21 slices (7 slices per breath-hold), FOV=400mm, and a 30-channel receiver phased-array RF coil. Streaks from the arms were the dominant source of artifact in this dataset and ROIs to isolate the left and right arms were used to create the interference matrix. The B-STAR method is compared with an automatic coil removal3 method (CR) and with the sensitivity weighted approach2 (SW). Representative results for composite images created from combining data from all 32 echoes are shown in Figure 3. Note that the SW method can handle mild levels of streaking but is unable to suppress strong streaking. The CR method performs well in removing streaks but exhibits a noticeable loss in signal levels. The B-STAR approach in contrast offers a better balance between artifact removal and signal retention. Quantitative metrics1 averaged across subjects/slices and ROIs demonstrates the performance of the different methods.
The multi-contrast data from the radFSE sequence can be reconstructed to yield TE images which can be fit to a T2 map. Figures 4 and 5 presents the results of view-sharing13 driven multi-contrast imaging and T2 mapping. Representative TE images and T2 maps from ACC, CR and the B-STAR methods are presented. Note the presence of streaking artifact in the ACC results (more noticeable in the T2 map as highlighted) and signal loss (along with increased noise amplification) in the CR results. In contrast, the B-STAR T2 map does not have residual streaking or loss of signal.
Conclusion
We present an extension to the B-STAR approach to suppress radial streaking artifacts in multi-contrast imaging and parameter mapping applications. Since the proposed method relies on the coil combination step for artifact suppression, it is a post-processing step and can be easily integrated into existing processing pipelines.Acknowledgements
No acknowledgement found.References
1. Xue Y, Yu J, Kang HS, Englander S, Rosen MA, Song HK. Automatic coil selection for streak artifact reduction in radial MRI. Magn Reson Med 2012;67: 470–476.
2. Kholmovski EG, Parker DL, Di Bella EV. Streak artifact suppression in multi-coil MRI with radial sampling. In Proceedings of the 15th Annual Meeting of ISMRM, Berlin, 2007. p. 1902.
3. Grimm R, Forman C, Hutter J, Kiefer B, Hornegger J, Block T. Fast automatic coil selection for radial stack-of-stars GRE imaging. In Proceedings of the 21th Annual Meeting of ISMRM, Salt Lake City, 2013. p. 3786.
4. Holme HC, Frahm J. Sinogram-based coil selection for streak artifact reduction in undersampled radial real-time magnetic resonance imaging. Quant Imaging Med Surg 2016;6: 552–556.
5. Feng L, Chandrana H, Sodickson DK, Otazo R. Unstreaking: Radial MRI with automatic streaking artifact reduction. In Proceedings of the 25th Annual Meeting of ISMRM, Honolulu, 2017. p. 4001.
6. Feng L, Huang C, Shanbhogue K, Sodickson DK, Chandarana H, Otazo R.RACER-GRASP: Respiratory-weighted, aortic contrast enhancement-guided and coil-unstreaking golden angle radial sparse MRI. Magn Reson Med 2018;80: 77–89
7. Block KT, Fenchel M.Simple Method for attenuation of streaking artifacts from peripheral intensity accumulation.In Proceedings of the 15th Annual Meeting of ISMRM, Melbourne, 2012. p. 2444.
8. Mandava S, Keerthivasan MB, Martin DR, Altbach MI, Bilgin A. Radial streak artifact reduction using phased array beamforming. In Proceedings of the 26th Annual Meeting of ISMRM, Paris, 2018. Abstract No. 0933.
9. Walsh DO, Gmitro AF, Marcellin MW. Adaptive reconstruction of phased array MR imaging. Magn Reson Med 2000;43: 682–690.
10. Van Veen BD, Buckley KM. Beamforming: A versatile approach to spatial filtering. IEEE ASSP magazine 1988; 5: 4-24.
11. Kellman P, McVeigh ER. Ghost artifact cancellation with phased array processing. Magn Reson Med 2001;46: 335–343.
12. Monzingo, Robert A., and Thomas W. Miller. Introduction to adaptive arrays. Scitech publishing, 2004
13. Altbach MI, Bilgin A, Li Z, Clarkson EW, Trouard TP, Gmitro AF. Processing of radial fast spin-echo data for obtaining T2 estimates from a single k-space data set. Magn Reson Med 2005;54:549–559.
Figures
