Respiratory self-navigation (SN) in 3D radial segmented MR imaging [1,2] is an efficient and robust method for the depiction of coronary luminal anatomy using MRI when compared to more standard navigator-gated acquisitions [3]. This method has largely been used at 1.5T in combination with bSSFP readouts, both with and without gadolinium-based contrast agent injection, to provide the high blood-to-myocardium contrast required for vessel segmentation and respiratory motion tracking [4,5]. However, at higher magnetic field strengths, bSSFP imaging is often confounded by off-resonance artifacts and conventional gadolinium-based contrast agents combined with inversion-recovery preparation and gradient-echo readouts often cannot achieve a constant signal throughout the scan. Variable contrast can cause deterioration to the tracking signal of self-navigation and therefore to the resulting motion correction. Here we investigate the use of an intravascular contrast agent based on ultra-small super-paramagnetic iron oxide particles (USPIO) with longer half-life as a potential method to improve the performance of self-navigation at 3T.In this preliminary study, N=10 healthy volunteers were recruited in accordance with local ethical approval. Segmented, 3D ECG-triggered, respiratory self-navigated, whole-heart coronary MRA was performed. All datasets were acquired using a prototype 3D radial sequence [6] on a 3T clinical system (MAGNETOM Verio, Siemens Healthcare, Erlangen, Germany). Using an empirically selected inversion time of 300ms good myocardial suppression and appropriate enhancement of the blood pool was obtained over a 40-50min window (allowing repeat imaging if required). Sequence parameters were: TR/TE 3.5/1.7ms, FOV (220)mm³, matrix 192³, voxel size (1.15mm)³, RF excitation angle 15° and receiver bandwidth 1000Hz/Px. About 12000 radial readouts were acquired in 300-400 heartbeats, with an overall undersampling ratio of 20%. Intravenous ferumoxytol (510mg of elemental iron (Fe) in 17 ml (30 mg/ml)) at a dose of 4mg of iron (Fe)/kg of body weight was administered over a 15min infusion (in accordance with international guidelines). In all cases both diastolic and systolic imaging was undertaken and evaluated for reliability of the self-navigation signal and final image quality. Expert assessment of an experienced radiologist (SM), who graded the Aorta (Ao), the main pulmonary artery (PA) and vein (PV), the superior (SVC) and inferior (IVC) vena cava, and the proximal- and mid-left anterior ascending (LAD) and right coronary arteries (RCA) was performed on a 5-point scale: 0 = not visible, 1 = markedly blurred, 2 = moderately blurred, 3 = mildly blurred and 4 = sharply defined. Finally, coronary vessel sharpness and length were computed as described in [7].In almost all cases both the SN signal and the final image quality were good (Fig.1,2) –with the only two confounding issues observed in individual cases, 1) sub-optimal fat-saturation reducing CNR in small vessels and 2) large non periodic motion (heavy inspirations) in subjects with less regular respiratory pattern. The long half-life of the contrast agent enabled an identical inversion time to be used for both systolic and diastolic acquisitions, providing good tissue contrast for both datasets. The tracking was successful in all 20 acquisitions. The grades from the expert assessment were very similar between diastolic and systolic acquisitions for the great vessels with averages of: 3.7±0.4 (Ao), 3.6±0.5 (PA), 3.3±0.5 (PV), 3.6±0.5 (SVC), and 3.5±0.5 (IVC). The average grades, sharpness, and length for all evaluated coronary segments are reported in Fig.3. All reported values did not show statistically significant differences between diastolic and systolic datasets, although a trend for increased sharpness of the systolic RCA and length of the diastolic RCA could be noticed (Fig.3,4).The long-lasting contrast enhancement provides a constant signal enhancement in both spatial and temporal dimensions and enables high quality MRA acquisitions. The very high blood-to-background contrast is ideal for the tracking of the blood pool performed by the self-navigation algorithm, both for diastolic and systolic acquisitions. Although trends for differences in the delineation of the RCA can be noticed, no statistical significance was reached. Potential improvements include methods for automated rejection of data acquired during anomalous respiratory motion [8] and further optimization of image contrast. The longer acting contrast agent allows additional data to be re-acquired in problematic cases, while volume rendering of the 3D datasets becomes easier due to the high contrast of the vessel lumens (Fig.5). As USPIOs can be used as biomarker for inflammation there is the potential to use this methodology as both an angiographic method and as a method for detection of concentration of USPIO in areas of inflammation.We have demonstrated the feasibility of T1 shortening USPIO based contrast enhanced respiratory self-navigation for whole-heart free-breathing coronary MRA at 3T for both diastolic and systolic acquisitions.