Self-navigated 3D late gadolinium enhancement imaging of the left atrium

Introduction: Late Gadolinium Enhancement imaging is evolving as a valuable tool for treatment management of atrial fibrillation using radio-frequency ablation [1, 2]. Enhancement in images post-ablation correlates with regions of ablation [1, 2] that can be used to track treatment progress and identify gaps in the ablated regions. In pre-ablation images, enhancement corresponds to fibrosis which could be a useful tool in predicting the success of the ablation procedure [3]. The current standard for obtaining LGE images is to use a respiratory navigated 3D Cartesian acquisition with parallel imaging [1, 2]. Segments of 3D k-space are acquired each heartbeat in the diastolic phase of the heart with ECG gating and those within a navigator window of motion of the diaphragm are kept while others are discarded [2, 3]. While good quality images can be obtained using this approach, it has been reported that ~20-40% of the scans are non usable due to poor quality [3, 4]. Motion in the images can play a significant role in determining the image quality and depends on the breathing pattern of the patient. The acquisition time can also be long. Here we propose an alternative self-navigated hybrid radial scheme that can overcome some of these limitations. We exploit the self-navigation property of a radial acquisition along with its robustness to undersampling to obtain good quality images efficiently. In addition, we use a total variation based compressed sensing reconstruction to improve the image quality from undersampled data.