Towards robust c-spine imaging with Cartesian sampling
Guobin Li1, Zhaopeng Li1, Chaohong Wang1, Yang Xin1, Shuheng Zhang1, Weijun Zhang1, Xiaodong Zhou1, and Weiguo Zhang1

1Shanghai United Imaging Healthcare Co., Ltd, Shanghai, China, People's Republic of


C-spine imaging is demanding due to artifacts from CSF flow, patient’s swallowing etc., especially in FSE sequence with inversion recovery. By increasing the excitation thickness in FSE sequence, two concomitant saturation bands are realized at both sides of each image slice, which suppress the moving CSF. Furthermore, a snapshot k-space ordering is proposed to further improve the stability of c-spine imaging against irregular flow of CSF and patient’s bulk motion.


MR imaging of c-spine is often impaired by the motion artifacts from CSF, especially in fast spin echo (FSE) sequence with inversion recovery. As revealed in previous studies, CSF has very complex flow patterns affected by cardiac, respiratory, and other factors1,2. These problems make it very demanding to obtain stable c-spine images with Cartesian sampling. Non-Cartesian acquisitions can be used to address these problems, but they sacrifice sampling efficiency and contrast flexibility3. In this work, methods are demonstrated for obtaining robust and optimal c-spine images by a combination of several simple techniques in FSE sequence with regular Cartesian sampling.


The proposed solution combines two techniques: (1) concomitant saturation bands; (2) snapshot k-ordering. In the first technique, the gap between two adjacent slices is made equal to or greater than the slice thickness. The excitation thickness is greater than the slice thickness, and the refocusing thickness is equal to the slice thickness. In-flow CSF sees the excitation pulse first, but no refocusing pulses, therefore it accumulates de-phasing from encoding and crushing gradients before arriving at the imaged plane. FSE sequence evolves into two concomitant saturation bands at both sides of each imaged slice, which suppress any rapidly flowing CSF, and reduce data inconsistency. Full spatial coverage is achieved by shifting the position of excited slices to un-acquired regions after the acquisition of a slice group. Though the concomitant saturation bands can significantly reduce the data inconsistency due to CSF flow, they cannot totally avoid such occurrence of in extreme cases such as when central k-space data are acquired at the peak flow of CSF ( e.g. in systolic phase), or when patient swallows. To further improve the success rate of such examination, a novel k-space view ordering is proposed (referred to as snapshot k-ordering). As demonstrated in Figure 2, data from a train of N echoes are sequentially filled into N adjacent k-space locations. A subsequent echo train is filled into the next N adjacent k-space locations with an inversed order and the procedure is repeated until all data are collected. The periodic modulation of the signal intensity due to T2 attenuation can be alleviated by averaging. In case of two averages, for example, all data are acquired twice with inversed order, and averaged during reconstruction. Alternatively, variable refocusing flip angles can be utilized to achieve smooth signal evolution over the echo train4. The proposed solution was implemented into a 3T clinical MR scanner. Over 20 volunteers were imaged with consent. For comparison, conventional and proposed FSE imaging were interleaved for each examination. Inversion recovery was used for fat suppression in all cases.


In the image acquired with conventional FSE sequence (Figure 3a), both the vertebra and the spinal cord were often seriously contaminated by pulsatile artifacts of CSF. In contrast, CSF artifacts were significantly reduced in the image acquired with the proposed FSE sequence with concomitant saturation bands (Figure 3b). The advantage of the snapshot k-ordering is its robustness against volunteers’ uncooperative behaviors, e.g. bulk motion, swallowing etc. Figure 4 shows the difference in image quality between the conventional and proposed FSE imaging when the volunteer became impatient and mobile.


By employing concomitant saturation bands, artifacts of rapidly moving CSF can be significantly reduced. Compared to conventional saturation band, concomitant saturation band is superior because it does not lengthen the minimum TR, and has not gap between the saturation band and the imaged slice. Much faster sampling of k-space central region is achieved with the snapshot k-ordering, which explains why the image quality of snapshot k-ordering was always more stable in the interleaved comparison with the conventional k-ordering. By combining these techniques, c-spine imaging with reliably high quality is made feasible even with normal Cartesian k-space sampling.


No acknowledgement found.


1. Yamada S. et al. Am J Neuroradiol 36:623–30, 2015

2. Yamada S. et al. Fluids and Barriers of the CNS 2013, 10:36

3. Fellner C. et al. Am J Neuroradiol 31: 674-81, 2010

4. Busse R.F. et al. Magn Reson Med 60:640–649,2008


Fig1. FSE sequence evolves into two concomitant saturation bands at both sides of each imaged slice by increasing only the excitation thickness. Any in-flow CSF or blood signals will be suppressed first by the concomitant saturation bands.

Fig2. Snapshot K-Ordering. Each shot sequentially fills multiple neighbouring k-space locations. All data are acquired twice with inversed filling order and averaged to alleviate T2 attenuation.

Fig3. Suppression of pulsatile artifacts of CSF using concomitant saturation bands. Left: conventional FSE; right: proposed FSE with increased excitation thickness.

Fig4. Variation of image quality in the scanning of an uncooperative volunteer with conventional FSE (#1 and #3) and proposed FSE (#2 and #4). The number in the lower left of each image indicates the acquisition order.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)