This prototype, self navigated, free breathing 3D sequence (SNFB3D) [1] acquires continuous ECG-segmented radial views of the heart with 100% scan efficiency and without the need of a diaphragmatic beam-navigator [2]. This negates the time expense of obtaining cardiac images only within a range of diaphragmatic positions. The acquisition follows a spiral phyllotaxis pattern for the distribution of the 3D radial readouts [3]. The reconstructed image is respiratory motion corrected based on the automated detection of the inferior-superior motion of the left ventricular blood pool. Acquisition times are typically 5-6 minutes and the duration of the scan is known prior to scanning.This is a retrospective review to determine the sensitivity of the SNFB3D in diagnosing coronary artery origins. Additional grading of the images was performed to quantify the quality of images obtained and to attempt to describe some factors affecting image quality.Our pediatric cardiac MR team performs all pediatric, and almost all, adult congenital MRs obtained at our center. Studies performed between 2/14 and 7/15 were reviewed. Self-navigated 3D datasets were acquired in 109 studies in 107 patients (average age 20 years, range 0.1 to 58.5 years). Protocol parameters of the fat-saturated, T2-prepared imaging sequence were set as follow: TR/TE 3.1/1.56ms, FOV (220mm)³, matrix 1923, voxel size (1.15mm)³, RF excitation angle 90°, and receiver bandwidth 898Hz/pixel. Studies were reviewed by a single pediatric cardiologist. For a scan to be considered diagnostic it must unambiguously display the origins of the left main (LMC), left anterior descending (LAD), and right coronary (RCA) arteries. Diagnostic quality was subjectively graded based on an adaptation of a previously described grading scale [4] (1 = non-diagnostic, 2 = sufficient for diagnosis but with considerable blurring and reduced vessel sharpness, 3 = good quality with minor blurring and mildly reduced vessel border sharpness, and 4 = excellent quality without significant blurring and with sharp vessel borders). The ability of the sequence to freeze cardiac and respiratory motion was subjectively graded (1= motion affecting diagnosis, 2 = motion but not affecting diagnosis, 3 = no significant motion). Respiratory motion was graded based on the sharpness of the dome of the diaphragm and the internal mammary arteries. Cardiac motion was based on the sharpness of cardiac structures. Homogeneity of the intracardiac blood pool was subjectively graded (1 = inhomogeneity affecting diagnosis to 5 = no significant inhomogeneity). The ability to identify the origins of the RCA, LMC, LAD, circumflex (Cx), first diagonal (DIA), and posterior descending (PD) coronaries in addition to coronary dominance (DOM) was determined. Chi square or Fisher’s exact test were used to determine a relationship between a non-diagnostic scan and cardiac or respiratory motion, or inhomogeneous cardiac blood pool.A diagnostic study was obtained in 80.7% of scans. Of diagnostic scans, image quality was 33.0% sufficient (grade 2), 44.3% good (grade 3), and 22.7% excellent (grade 4). Coronary dominance could be determined in 57.8% of scans. Ability to identify the origin of individual coronary segments was LMC - 87.2%, LAD - 84.4%, RCA - 84.4%, CX - 79.8%, PD - 53.2%, and DIA - 28.4%. Percentages of scans falling within each subjective category for cardiac and repsiratory motion and for blood pool inhomogeneity are presented in figure 1. Obvious blood pool inhomogeneity was common but diagnosis was possibly affected in only 6 scans. The greatest factor affecting diagnostic ability was cardiac motion (P<0.01).Self-navigated 3D sequences can provide excellent sensitivity in diagnosing coronary origins with significant time savings compared to diaphragm navigated sequences. The ability to freeze cardiac motion remains a major determinant to image quality and diagnostic sensitivity.