We developed a stimulated echo based 3D diffusion-prepared turbo spin echo sequence (DPsti-TSE) to achieve high-resolution diffusion-weighted ‎imaging (DWI) of the carotid vessel wall. The sequence is robust to eddy currents, the presence of pulsatile tissue motion, and B₁-inhomogeneity. DWI of the carotid is promising for detecting a lipid necrotic core and hemorrhage, without requiring contract agent injection [1,2]. In a previous study, a diffusion-prepared TSE (DP-TSE) sequence was applied to obtain ADC maps of the carotid vessel wall [1]. However, the DP-TSE sequence is very sensitive to eddy currents induced by the diffusion gradients. Even on modern MRI scanners induction of eddy currents cannot be avoided. Additionally, pulsatile motion of tissue and B1 inhomogeneity in the head-neck region lead to image blurring, and inaccurate and poorly reproducible ADC values [3]. In the present work we show robustness of the DPsti-TSE sequence to above-mentioned issues, as demonstrated by phantoms and in vivo scans of human carotid arteries. Sequence design: The DPsti-TSE sequence consists of a diffusion-preparation module and a TSE readout module (Fig.1). In the diffusion-preparation module, the magnetization is tipped into transverse plane to experience diffusion gradients and then restored in longitudinal direction for TSE readout module. To avoid additional non-diffusion-related signal loss caused by eddy currents, a dephasing gradient is placed before the tip-up pulse in slice encoding direction to evenly disperse the magnetization. Corresponding rephasing gradients are placed in the TSE module to recall the diffusion-prepared magnetization as stimulated echoes, while satisfying the CPMG condition. The area of dephasing/rephasing gradient is larger than the combined slice encoding gradients in the TSE readout to prevent stimulated echo artifacts. Four hard refocusing pulses are used in an MLEV pattern to decrease sensitivity to B₁-inhomogeneities [4]. In between, m1 nulling diffusion gradients are used to decrease bulk motion sensitivity. Velocity encoding (Venc) gradients can also be added to introduce a moderate m1 value, thereby suppressing flowing blood. All scans were performed with a Philips 3T scanner (Ingenia, Philips). Phantom measurement: The proposed DPsti-TSE sequence was compared with DP-TSE sequence in a homogeneous phantom. Three b-values were applied in x, y and z directions in scanner coordinates to obtain ADC maps. All sequence parameters were identical. ADC values were also derived from conventional echo planar imaging (EPI) DWI sequence for reference. In vivo measurement: Two healthy volunteers (50y-M and 22y-F) were scanned by the DPsti-TSE and DP-TSE sequences, respectively. Two b-values were used in (1, 1, 1) direction in scanner coordinates to maximize the gradient strength to 72mT/m. Diffusion-preparation duration was 47ms. Images were recorded at end-diastole using cardiac triggering. Venc gradients selection velocity was 1.5cm/s. Other sequence parameters were: TSE factor=16, TE=28ms, echo spacing=4.0ms, TR=2R-R, voxel size=0.6×0.6×3mm³, field-of-view (FOV)=130×48×39mm³. Oversampling was used for fold-over suppression in anterior-posterior direction. Total scan time was 11min52s. The DPsti-TSE sequence generated identical ADC maps in three directions with values equal to those obtained with EPI. (Fig.2). In contrast, DP-TSE sequence yielded largely overestimated ADC values in x-direction and y-direction 10- and 6-fold, respectively. The ADC value in z-direction was slightly higher than the reference. These observations imply that the eddy currents were strong when diffusion gradients were placed in x-direction and y-directions leading to significant additional signal loss. In vivo, transversal DP-TSE images of the neck region with carotid arteries also displayed non-diffusion related signal decay and blurring, with SNR dropping below acceptable level for b-value=300s/mm² (Fig.3). The new DPsti-TSE sequence performed significantly better. The carotid vessel wall can be delineated with good anatomical definition and efficient blood suppression in the lumen. ADC values calculated in the vessel wall and surrounding muscle were (1.4±0.3)×10^-3mm²/s and (1.2±0.4)×10^-3mm²/s, respectively. Phantom scans demonstrated that DPsti-TSE sequence provided correct ADC estimations in the presence of eddy currents. In vivo scan showed that DPsti-TSE sequence could provide high-resolution vessel wall images with good blood suppression and wall visualization. The ADC values in the wall obtained from DPsti-TSE sequence agreed with previous studies [1,2]. Some minor blurring can be observed at b-value=300s/mm², indicating that some residual motion sensitivity. Carefully tuning the cardiac trigger delay could further mitigate the motion interference. Vessel wall delineation could be further improved by optimizing the TSE flip-angle sweep to increase image sharpness [5].Our proposed DPsti-TSE sequence proved to be robust against eddy current artifacts, motion and B₁-inhomogeneities and allows for sub-millimeter resolution diffusion-weighted carotid vessel wall imaging.