Advanced computed tomography (CT) allows for amazing visualization of the human body including the beating heart. However, the complex interactions of blood flow, which is crucial in the diagnosis and treatment planning of many diseases, are not fully reflected by these images. Magnetic resonance imaging (MRI) and ultrasound are able to measure functional data like blood flow, but at a low resolution. Furthermore, these techniques are not able to predict the changes in blood flow after surgical treatment.
This project aims to extract blood flow data from CT images of the heart using image-based simulations. The goal is earlier and more accurate detection as well as improved management of cardiac diseases.
From Simple Models to In Vivo Measurements
Even though many forms of functional imaging data and modelling approaches are currently available, a gap persists between modelling and experimental research. This project has bridged the gap by developing and evaluating an approach in which intracardiac flow fields are computed based on patient-specific high-resolution cardiac CT data. The heart is segmented and advanced registration techniques are used to track the heart wall. Using computational techniques usually employed by the automotive or aerospace industry, detailed intracardiac and vascular blood fields are obtained.
The results show that the 4D Flow CT method can produce blood-flow patterns that are qualitatively and quantitatively similar to the current reference standard 4D Flow MRI, but at higher resolution. The high resolution also allows the simulated data to reveal processes that couldn’t be studied before, like the coagulation of blood or the occurrence of turbulence in the blood flow.
One clinical application that is explored is in atrial fibrillation. These patients have an increased risk of blood cloths forming in the atrium and by migrating to the brain or coronary arteries they may induce a stroke or heart attack. We are building a model that can identify where the blood cloths are forming. The goal is that the information from this model may be used to identify patients at risk.
The simulation-based approach potentially allows for studies of what-if scenarios where different treatment options can be explored. This is challenging, as the heart is complex and adapts to changes in demand and constrains. As a model is a simplified version of reality, there has to be a balance in the amount of details included and clinically usability.