Congenital heart disease (CHD) is the most common birth defect, affecting 1.2% of all live births. Imaging plays a major role in managing CHD but remains challenging for evaluating complex cardiac and vascular abnormalities across a wide range of age and habitus. To address these limitations, the PIs have developed cardiovascular 4D flow MRI which can measure complex 3D blood flow in-vivo, a task difficult or impossible to obtain with other imaging strategies. Recent efforts have focused on two forms of CHD: 1) bicuspid aortic valve (BAV) which is the most common form of CHD, and 2) single ventricle physiology (SVP), one of the most severe forms of CHD. Our 4D flow MRI studies have successfully identified new hemodynamic biomarkers to better characterize CHD. We were the first to establish a physiologic link between aberrant 3D blood flow, elevated wall shear stress (WSS), aortopathy phenotype, and aortic wall tissue degeneration on histopathology in patients with BAV. In patients with SVP, our findings demonstrated relationships between surgical correction strategies and flow distribution to the lungs, a known factor implicated in SVP outcome. We have achieved successful clinical translation at Northwestern, where 4D flow MRI is now used as a clinical tool in diagnostic MRI exams for patients with CHD and aortic disease. Over the past four years, the PIs have assembled one of the largest 4D MRI databases with over 2500 patient exams. For this renewal application, we identified a need to increase the dynamic range of 4D MRI flow sensitivity to account for data complexity (3D + time) and the wide age range in CHD by a combination of dual-venc flow encoding, compressed sensing, and SSFP imaging. Second, three is a need for longitudinal studies to identify predictors of BAV and SVP outcome. Third, making these unique but complex 4D MRI data sets and analysis tools more widely available to the greater research community is challenging. In addition, no automated methods currently exist for advanced processing such as atlas based analysis across large cohorts. Analysis is thus time consuming and requires manual interactions (e.g. 3D vessel segmentation) which limits reproducibility and translation. To address this need, an established Northwestern data archival and pipeline processing resource based on remote high-performance computing clusters (NUNDA) will be utilized for standardized data archival, sharing, and pipeline processing of 4D MRI data. This platform will provide the unique opportunity to utilize annotated data available in the 4D MRI database (>1300 BAV, SVP, and control 4D MRI data analyzed in the initial funding cycle) for application of machine learning concepts to establish (semi-)automated 4D MRI analysis workflows in NUNDA. Thus, the renewal application for this study aims to 1) develop a rapid (15 min) non-contrast 4D MRI for clinical translation, 2) leverage the existing large 4D MRI database to identify 4D MRI metrics predictive of long-term (> 5 years) CHD patient outcome, and 3) establish a remote NUNDA platform for 4D MRI data sharing and automated analysis across large cohorts.
Our goal is to develop non-contrast 4D MRI, a new diagnostic test to achieve an improved assessment for the most common and one of the most severe forms of congenital heart disease: bicuspid aortic valve and single ventricle physiology. We will leverage an existing large 4D MRI database to allow for long-term 5-8-year follow-up to establish new measures for improved outcome prediction and therapy management for patients with bicuspid aortic valve and single ventricle physiology. A comprehensive data archive will be established to allow for the dissemination of the 4D MRI data, analysis tools, and study results to the greater research community.
Showing the most recent 10 out of 78 publications
