Half-Fourier single shot spin echo sequence (HASTE) is a turbo spin echo (TSE) technique and uses a single-shot method to acquire an entire 2D image in a single TR. Multi-slice HASTE is widely used in the abdominal imaging with breath hold due to its short scan time. However the imperfect slice profile of the selective RF pulse will lead to the undesired excitation of the neighboring regions. The adjacent slice will be partially saturated, leading to signal attenuation along the slice direction. Unless a very long TR time is used, crosstalk between slices could cause artifacts or SNR loss in 2D acquisitions. One conventional solution for this problem is to modify the slice acquisition order and increase the effective gap between sequentially excited slices¹. In addition, the RF pulse train can be considered as off-resonance magnetization transfer (MT) pulses for other slices, resulting in signal attenuation and contrast alteration. The MT effect could also be reduced by increasing the slice gap and altering the slice order^2-3. However the slice gap should not be increased unlimitedly because small lesions can be missed. The crosstalk and MT induced artifacts could not be eliminated by simply acquiring slices with odd/even ordering, in particularly for the imaging with thinner slices. In this study, an optimized slice acquisition ordering method was proposed to improve the signal attenuation and contrast alteration in multi-slice HASTE imaging using a short TR.

Theoretically, the longer interval between excitation pulses is preferred for the neighboring slices. In order to maximize the slice gap between successively excited slices and interval time between excitation pulses of neighboring slices, the acquisition of neighboring slices could be interleaved iteratively. Take 20 slices with 2 concatenations as an example, illustrated in Fig.1. The slices represented by squares are numbered based on its location in the imaging volume. Different acquisition orders are depicted using dot lines with different colors. The conventional odd/even slice acquisition order is indicated with light gray squares and yellow dot line. In the proposed method, the slices are divided into two groups at first, in which odd slices are acquired followed by even slices. Then slices within each group will be interleaved iteratively, shown with gray squares and green line. In Fig.1, the slices can be interleaved three times at most if there is no criterion, represented by dark squares and red line. However this acquisition order is not the optimized one, e.g. the interval is too short between #13 and #11 slices. Therefore delicate criteria are needed to avoid excessive iterations. In this study, we assume that the signal attenuation is negligible when the slice gap of successive excited slices is larger than 3 times of slice thickness or the excitation interval neighboring slices is longer than 2 second. Although the diagram only illustrates the situation for the slice number of 20 with 2 concatenations, the same principle is also applicable to the other slice numbers with different concatenations.

The proposed method was implemented in HASTE sequence and validated via in-vivo study. Abdominal images were acquired from a healthy volunteer using a commercial 1.5T scanner (MAGNETOM Aera, Siemens Healthcare). Conventional and non-product HASTE were acquired with same parameters respectively. The imaging parameters are as follows: TE/TR = 83/(1300, 800, 600 and 400) ms, slice thickness = 6 mm, 20% slice gap, matrix size = 198×256, slice number = 30, concatenation = 2, in-plane GRAPPA factor = 2.

Fig.2 shows abdominal images acquired with conventional and non-product HASTE sequence respectively. Fig.2a is acquired with original HASTE sequence, TR = 1300ms. Fig.2b-d are acquired with modified HASTE sequence, TR = 800ms, 600ms and 400ms respectively. For these images, it notes that the SNR and contrast in Fig.3b-d are very similar to that of Fig.2a, even with one third of TR time (Fig.2d). Fig.3 shows the corresponding 3D reconstructed abdominal images. All images are showed with same windowing values. In comparison to images acquired with conventional odd/even slice acquisition ordering, cerebrospinal fluid (CSF) signal is more homogenous along the slice direction in Fig.3b,d-f, even with a shorter TR.It demonstrated that a shorter acquisition time is possible with an optimized slice acquisition order. It allows a shorter TR, whilst maintaining the SNR and contrast similar to the conventional one, which is particularly useful in the abdominal imaging. The proposed slice acquisition ordering method could be widely used in HASTE sequence. In addition, it could also be easily adapted to the other multi-slice imaging sequences.