Cardiovascular magnetic resonance (CMR) late gadolinium enhancement (LGE) is the clinical gold standard for assessment of myocardial viability [1]. Commonly, a series of 2D slices in short axis and long axis views are used to cover the entire left ventricle. This 2D approach requires multiple breath-holds and limits the imaging spatial resolution, especially in through plane (8-10mm). 3D LGE imaging has been developed as an alternative to 2D LGE to simplify data acquisition and increase spatial resolution of 3D LGE. Furthermore, 3D LGE enables assessment of complex 3D scar geometry, which is essential in determining the substrate for ventricular arrhythmia [2,3]. Despite the potential of 3D LGE, the scan time remains long, which hindered its clinical adoption. We have recently developed 3D LGE imaging using compressed-sensing based reconstruction of LOST [4,5]. Additionally, we have integrated reconstruction workflow into our clinical scanner to facilitate 3D LGE imaging in a clinical setting [6]. In this study, we sought to investigate the image quality of 3D high-resolution LGE in patients undergoing clinical CMR and study the impact of patient characteristics on image quality.To perform visual quality assessment of high resolution LGE with isotropic spatial resolution of 1-1.5mm³ acquired using LOST with acceleration factors of 4-5 and to determine the impact of patient characteristics on image quality.In a prospective study, we recruited 268 patients with known or suspected cardiovascular disease undergoing clinical CMR over a period of 53 months. All subjects were imaged using a 1.5T MR scanner (Achieva; Philips Healthcare, Best, the Netherlands) and a 32-channel cardiac phased array receiver coil. The study was Health Insurance Portability and Accountability Act (HIPAA) compliant, and the imaging protocol was approved by our institutional review board. 3D LGE sequence was performed 10-25 min after administration of 0.1-0.2 mmol/kg of Gd-DTPA or Gd-BOPTA. The typical imaging parameters were as follows: segmented bSSFP imaging readout, TR/TE=6.1/2.7ms, flip angle=25, FOV=320x376x112mm³, voxel size= 1mm³ to 1.5 mm³, TFE shots=482, TFE factor=22, acquisition window=134.7 ms, low-high k-space ordering, 5 linear ramp-up pulses. A respiratory navigator placed on the right hemidiaghram was used for gating and tracking. The typical scan time was 6 minutes 28 seconds (1 mm³ voxel size, acceleration rate=5) or 1 minute 54 seconds (1.5 mm³ voxel size, acceleration rate=5) for a gating efficiency of 100%. All images were acquired using axial orientation covering the entire heart. A 3D randomly undersampled acquisition sequence was implemented for the accelerated acquisitions, and a net acceleration rate of 4-5 in ky-kz was used [7]. All images were reconstructed automatically using LOST reconstruction on the scanner and stored in hospital PACS. Image quality was assessed subjectively by a radiologist, blinded to the reconstruction method, using 1-4 scale, with higher scores indicating better overall image quality. Presence of hyperenhancement was assessed in each patient. Description of artifacts encountered were recorded for analysis. Several patient parameters were analyzed: age, gender, heart rate, weight, patient’s anteroposterior (back to chest wall) and transverse (right to left arm) dimension. To assess correlation between visual quality assessment scores and various patient factors, logistic regression analysis was performed. All statistical analysis was performed in SPSS v19.0 (IBM Corp, Armonk, NY).Figure 1 shows example LGE data set scored from 1 to 4. Overall average score was 3.2 +/- 0.8. Scores of 1 to 4 was given to 3.0% (n=8), 21.6% (n=58), 26.5% (n=71) and 48.9% (n=131) of patients. Scar was present in 32 of patients. Figure 2 shows an example case with subendocardial LGE and reformatted planes from axial 3D LGE. To assess the correlation between score and patient factors, data from those which was scored 1 was combined with those scored at 2, because of low number of patients scored at 1 (n = 8). There was a statistically significant correlation between several patient factors and visual quality assessment (VQA) scores (Figure 3). Increasing subject weight, anteroposterior dimension and heart rate negatively impacted scores: subject weight (p=0.02; PseudoR²=0.023), AP height (p<0.001; Pseudo R²=0.908), high heart rate (p=0.04; PseudoR²=0.017) (Table 1). There was no significant correlation between patient age or gender with the VQA score. 3D LGE with high isotropic spatial resolution (1-1.5 mm³ ) is clinically feasible and yield sufficient image quality in majority of patients (3.2 +/- 0.8) . Patient size (torso dimension and weight) and heart rate adversely impact image quality.