Biomedical Engineering

The human respiratory system is a very complex organ, which spans many different scales. From an imaging and/or modelling point of view, reconstructing of the airway, finite element discretisation and solution are very challenging problems. There is in fact a huge lack of knowledge on the respiratory system either as a whole, or considering the lungs and the airways separately.

We are currently involved in a collaborative project on a very specific topic, to investigate the clinical usefulness of using objective measurements, as a tool in the septoplasty (surgical procedure to correct deformities of the nasal partition between the nostrils, thought to contribute to nasal obstruction) decision-making process.

The use of CFD should provide a detailed airflow analysis of the complex conduit shapes of the nose and help to elucidate the aerodynamic significance of septoplasty.

The aortic valve, situated between the heart and the aorta, ensures a uni-directional flow and efficient pumping of the heart. If the opening and/or closing behaviour of this valve are compromising it can lead to serious complications or even death. We have developed fluid-structure interaction methods to capture the complex interactions between the blood flow and the valve leaflets. A geometric parameterisation of the valve leaflets allows us to investigate the influence of the valve geometry on the transvalvular flow and pressure gradients in a structured manner.

In another project we investigate the interactions the valve has with the heart and the arterial system. When a patient has multiple valve lesions the individual contributions might be obscured by the close interactions in the system. Mathematical models combined parameter estimation techniques are used to characterise healthy and diseased arterial systems. We envisage that these models will be used for diagnostic purposes in the future.

Although a tremendous amount of research is dedicated to the arterial system, the mechanical functioning of the lymphatic system is rather unexplored. In collaboration with Imperial College London and Texas A&M University we are developing computational models of lymphatic valves to gain an in-depth understanding of the opening and closing behaviour of these structures. An important aspect in this line of research is the experimental validation of the developed models (performed in Texas), since currently hardly any information is available for building a reliable model. The computational models will provide important information required to predict the overall response of a lymphatic network model (London), which can be used to study the pumping behaviour of this system in health and disease.

‌Although 3D CFD models are a great way to study local flow phenomena in the arterial system, 1D models can provide the bigger picture.  A one-dimensional arterial network model has been developed that includes the heart, aortic valve, coronary circulation and largest 60 arteries in the human body.

The model is capable of describing flow and pressure in the entire network as a function of time. Furthermore it captures the many forward and backward pressure wave reflections occurring in the arteries.  This model is used to simulate and study diseases such as aortic stenosis, coronary stenosis, aneurysms and hypertension.

Medical devices like ventricular assist devices (VAD) or oxygenators are often black boxes with very limited knowledge of what goes on inside. CFD in combination with microCT scans or more general models like Darcy can provide a diagnostic tool for these devices that is used to optimize their design. The effects of fibre distributions and non-Newtonian behaviour of blood are investigated through local flow fields and pressure drops. These are then extended with sophisticated diffusion models to capture the local and global oxygenation.

Epidemic spread in a network of populated centres has been intensively studied by many research groups. We have developed a new approach based on the small initial contagion (SIC) approximation, i.e. the share of infective individuals triggering the outbreak in every centre is small compared to its population.

The SIC approximation works well for highly populated centres, e.g. large cities that are coupled by the migration processes between them. Basic equations for the outbreak time and epidemic spread speed by travelling waves have been obtained. Direct numerical simulations for both deterministic and randomized systems have confirmed the analytical results.

The work is done in cooperation with the Department of Mathematics and College of Medicine. 

The endothelium consists of a single layer of cells lining the inner surface of blood vessels. Although we cannot characterise in details how endothelial cells in vivo respond to haemodynamic stimuli, and how these responses are influenced by mass transport phenomena, we know that they play an important role in maintaining healthy blood vessels.  

Using anatomically accurate geometric models of the blood vessel lumen,  together with structural models of the vessel walls, fluid-structure interaction strategies, rheological, cellular and transport models, we aim to develop a general framework for modelling the different mechanisms associated with blood flow regulation, to better understand the role of the endothelium in cardiovascular disease. In order to verify the resulting models, we are initiating different collaborative projects with experimental scientists.

The geometries of arterial ducts can vary greatly and have been linked to disease initiation and progression. In our group a framework has been developed that focuses on geometry reconstruction, meshing and CFD analysis for biological flows. Each of the softwares within this framework have been developed in our group.

The starting point for these studies is MRI/CT/ultrasound data from individual patients, which provides us with the geometries for the CFD studies. The framework allows us to simulate and analyse the flow in individuals and to investigate the effects of surgical interventions. The flow through various stenosed and healthy carotid geometries has been simulated and rigorous convergence studies have been performed. Also virtual corrections of an aortic aneurysm have been performed to study the flow behaviour before and after operation.‌

Segmentation of typical 3D CT or MR scans with the help of available commercial software is still cumbersome and time consuming. In cooperation with our Computer Science Department, a new robust and effective method has been developed which allows almost automatic segmentation of topologically complex objects from 3D and even 4D (3D + time) medical scans corrupted by noise, image artefacts, intensity homogeneity, etc.

This method, called the Geometrical Potential Force (GPF), belongs to a family of gradient-based active surface methods and is currently under further development to include the account for statistical shape information (priors). Another proposed segmentation method is based on Graph Cuts and it allows for a completely automated segmentation of blood vessels from Intra-vascular Ultrasound (IVUS) and optical coherence tomography (OCT) images. ‌