Improved diagnosis and treatment of congenital heart disease using simulation-based, ultra-resolution blood flow imagnig
Abstract
Roughly 1% of all children are born with heart defects, ranging from common and simple (ventricular and atrial septal defects) to more complex (e.g. tetralogy of Fallot). In severe cases, multiple surgeries are needed to save the newborn, resulting in a single-ventricle heart, where the inferior and superior vena cava are connected directly to the pulmonary circulation (Fontan circulation). Many of these patients develop heart failure early in life, likely due to increased venous pressure and decreased pulmonary flow.
It has been hypothesized that exercise-induced flow inefficiency due to turbulence in the Fontan circula¬tion can be an outcome predictor for this patient group. Testing this hypothesis requires 4D blood flow measurement at exercise, requiring high temporal resolution and short scan times to accurately depict the rapid physiological changes of the cardiovascular system. This rules out current turbulence MRI methods.
Furthermore, patients with coarctation of the aorta (stenotic aorta at birth) constitute another group that can benefit from ultra-resolution flow imaging. These patients are treated in the neonate pe¬riod with surgery, but around 30% develop early signs of re-coarctation and 15% need early new intervention. This analysis requires ultra-resolution blood flow imaging beyond the current state of the art, as the aorta is only 3-6 mm in diameter in these children. Further applications in congenital heart disease will be quantification of pressure maps, ki¬netic energy and turbulence in the right ventricle and pulmonary artery in Tetralogy of Fallot and atrial septal defect patients.
To address these issues, we have recently developed a framework for simulation-based, ultra-high resolution reconstruction of flow data from magnetic resonance imaging. The framework uses adjoint CFD optimization techniques to obtain the CFD solution that optimally matches sparsely sampled MRI flow data. Computational testing and validation in phantom setups using laser particle image velocimetry has demonstrated the potential of the method to both accelerate clinical scans, increase data quality and to extract hitherto unobtainable clinical parameters such as vessel wall shear stress and stenotic pressure drops.
This study aims to extend our framework to robustly handle in vivo patient data, and apply the framework to investigate the previously mentioned pathophysiological hypotheses with fast imaging. The resources applied for will first be used to run blood flow reconstruction optimizations on clinical cases (properly de-identified and pseudonymized), and to further test, improve and optimize the accuracy of the overall framework. The resulting data may be used to follow up Fontan patients and guide surgical decisions and avoid early onset of heart failure.
