To develop a 3D Stack of Stars (SOS) acquisition with ciné-ROCS for use in contrast-enhanced carotid plaque imaging and plaque component identification. Thin fibrous caps are believed to be vulnerable to rupture, often resulting in arterial thrombosis or plaque embolization.¹ Post-contrast T1 weighted imaging has been used to aid in the detection of the fibrous cap overlying the necrotic core.² However, the accuracy of current techniques is limited by the small size of the fibrous cap, patient motion, and imperfect blood flow suppression. This results in many post contrast scans being discarded as clinically uninterpretable. Cardiac pulsations of the carotid wall can be as great as 1mm, causing blurring of the wall morphology particularly in post contrast imaging. We have implemented 3D SOS with retrospective cardiac cycle gated compressed sensing (ciné-ROCS) reconstruction³ to minimize, characterize, and eliminate these artifacts in post-contrast imaging. Using a 3D turbo FLASH sequence capable of SOS trajectories, blood suppression was realized by using a DANTE preparation⁴ before each turbo-FLASH readout train. Following the DANTE prep and a fat saturation pulse, the same radial line for all partitions was acquired with a centric k-space ordering in the slice direction. IRB approval was obtained and twenty symptomatic patients were enrolled and underwent informed consent prior to imaging on a 3T MRI (Tim Trio, Siemens AG). A custom carotid coil was used. The parameters for 3D SOS were: coronal orientation, FOV=160x160 mm², isotropic dimension=0.73 mm³, TE/TR=2.5/8.0ms, 54 slices per slab, DANTE prep time=150ms. Scan time was 5 minutes with two averages. 3D SOS measurements were acquired before and after contrast administration. Pre- and Post-contrast 3D T1w SPACE with DANTE prep imaging were also acquired. All measurements were acquired on the coronal plane with the same special resolution. 3D SOS acquisitions were performed in conjunction with a cardiac signal. Each data line was sorted into eight phases of the cardiac cycle for ciné-ROCS reconstruction. The resulting sequence of under-sampled images was reconstructed using sliding window to complete the measurement data of each bin. Fully sampled (static) images were obtained using standard reconstruction methods. Axial images were obtained from multiplanar reformats. Figure 1 shows the five consecutive reformatted pre-(a) and post-(b) contrast images from 3D T1w SPACE acquisitions, static pre-(c) and post-(d) contrast reformatted images, and eight phases of ciné-ROCS reconstruction post-contrast images(e1-e8) from 3D SOS measurements. The wall delineation (blue arrows) and calcification (red asterisks) were clearly shown in the both of fully (a,b,c,d) and under- sampled (e1-e8) images. The ciné-ROCS reconstruction demonstrates the similar SNR shown in the static images. Both the static (d) and ciné-ROCS reconstruction images show clear plaque (green asterisks) or vessel wall movement (blue arrows) due to cardiac motion relative to 3D SPACE images (b). Figure 2 shows the static (a) and ciné-ROCS reconstruction (p1-p8) images from the post-contrast 3D SOS acquisition obtained from the other subject. Tiny plaque motions are shown in ciné-ROCS reconstruction images. As opposed to circumferential radial expansion/contraction, the ciné-ROCS images show random shear plaque movements (shown in green asterisks).To acquire fully sampled ciné images with prospective ECG gated acquisition substantially lengthens the scan time of an already long 3D sequence. Ciné-ROCS image reconstruction is possible from a standard 3D SOS acquisition if two or more measurement averages are obtained. The radial k-space sampling employed with the SOS trajectory improves the performance of under-sampled data reconstruction algorithms such as sliding window ⁵ or CS where each sample line captures high signal energy in the k-space centre and undersampling occurs in more than one direction.⁶ In this study the sliding window algorithm was used. Other reconstruction algorithms such as Temporally Constrained Reconstruction⁶ or Robust Principal Analysis⁷ will be evaluated. Future work will also investigate the clinical evaluation of cardiac induced post-contrast images using the proposed technique and its improvements of demonstration of the fine details of plaque components such as fibrous cap compared to the current techniques. Ciné-ROCS reconstruction from 3D SOS acquisition demonstrates both improved demonstration of fine plaque structures and vessel wall movement due to cardiac motion to the regular reconstruction and may help provide information on symptomatic plaque development and response to treatment.