Single-shot echo planar imaging (ss-EPI) is widely used for clinical diffusion weighted imaging (DWI), for its fast acquisition and motion insensitivity. However, ss-EPI suffers from geometric distortion caused by magnetic susceptibility differences, which has the potential to mask the pathology, especially when imaging slices around skull base or temporal lobes¹. Alsop has introduced a single-shot turbo spin echo (ss-TSE) sequence for DWI which is much less sensitive to the distortion and chemical shift artifacts in EPI². However, DWI using ss-TSE has the limitation in spatial resolution due to the image blurring caused by long echo train duration, and the requirement of single shot to avoid artifacts caused by motion-induced phase errors. In this work, a multi-shot Cartesian TSE DWI method with improved in-plane resolution is proposed, and SF-MUSE method^3,4 is applied to correct the ghost artifacts induced by shot-to-shot phase variations.

Pulse Sequence The non-CPMG Cartesian TSE sequence proposed by Alsop² is implemented as in Fig. 1. The multi-shot TSE (ms-TSE) phase encoding profile order is set to linear to reduce image blurring, and partial Fourier acquisition is used for shortening TE.

Phase Correction Since physiological motion encoded with diffusion gradients can introduce phase variations in different shots, phase correction is indispensable for multi-shot DWI scans. In this work, as no additional navigators are acquired, the 2D nonlinear phase information is calculated inherently and used for correction in image domain. A MUSE-based method, SF-MUSE^3,4 is implemented for shot-to-shot phase correction.

Experiments All scans were performed on a Philips 3.0T MRI scanner (Philips Healthcare, Best, The Netherlands) using an 8-channel head coil. Data were acquired from 3 healthy volunteers. All human studies were performed under IRB approval from our institution. For the ms-TSE DWI, FOV = 230 × 192 mm², in-plane resolution = 1.5 × 1.5 mm², slice thickness = 5 mm, acquisition matrix = 152 × 128, Phase encoding direction = right/left, 4 shots with 19 echoes per shot, partial Fourier factor = 0.6, echo train duration = 114 ms, TE/TR = 80/2300 ms, number of slices = 24, scan time = 4.5 min, diffusion preparation was applied in 3 orthogonal directions with b = 1000 s/mm², NSA = 4. For comparison, ss-TSE DWI (echo train duration = 402 ms), ss-EPI DWI with GRAPPA = 2 (effective echo spacing = 0.416 ms), and 6-shot EPI DWI (effective echo spacing = 0.143 ms) with in-plane resolution = 1.5 × 1.5 mm² were also scanned for comparison. The ms-TSE DWI images reconstructed with no phase correction and with 2D nonlinear SF-MUSE phase correction were compared. The ss-EPI DWI reconstructed with GRAPPA, and 6-shot EPI DWI reconstructed with SF-MUSE images were also compared for evaluation of the distortion reduction.

Fig. 2 shows the isotropic DWI of two representative slices using ssTSE (a), ms-TSE with no correction (b), ms-TSE with 2D phase correction using SF-MUSE (c), along with the ss-EPI (d) and 6s-EPI images (e). The phase error causes severe ghost artifacts if no correction is performed. The proposed 2D correction method can correct these ghost artifacts, and provide less image blurring compared with ss-TSE. ms-TSE with SF-MUSE correction provides similar contrast but distortion-free diffusion images compared with ss-EPI, and even 6-shot EPI (much shorter echo spacing than ss-EPI) scans (yellow arrow heads).

The scan time for ms-TSE DWI is longer compared with ss-EPI (1 min). Phase modulation and other methods^1,5,6 can also be used for higher SNR efficiency non-CPMG TSE DWI at the cost of higher SAR. The imaging time and echo train duration can be further reduced by replacing the TSE readout with GRASE. Compared with PROPELLER DWI methods^1,5, our proposed method is more efficient but limited in ultra-high resolution, which is the limitation of self-navigated MUSE-based methods. Additionally, ms-TSE DWI can also be a good candidate for imaging around metal implants.

The proposed multi-shot Cartesian TSE DWI method with SF-MUSE phase correction can provide diffusion images free from ghost artifacts and insensitive to susceptibility induced distortions.