Miguel O. Bernabeu

Research Fellow
Research Interests: 

I currently develop two lines of research involving image-based simulation of blood flow. First, simulation of intracranial aneurysm haemodynamics in collaboration with the National Hospital for Neurology and Neurosurgery, UK. My work featured in the New Scientist magazine (#2906, Feb 2013) and has been published in journals and international conferences. I have recently secured funding to perform a proof of concept deployment of a computational platform to assist neurosurgeons in the treatment of brain aneurysms. Second, I am interested in the integration of computational models of blood flow and vascular mechanobiology (fluid-solid-growth modelling) for the study of vascular remodelling during angiogenesis and retinal vascular disease (in collaboration with teams at Cancer Research UK and NHS Moorsfield Eye Hospital, respectively). Both applications share the need for integrative approaches that bring together experimental and computational models in order to unravel the mechanisms that govern the response of blood vessels to normal and abnormal blood flow conditions. To support this work, I am one of the lead developers of HemeLB, an open source, parallel, lattice-Boltzmann blood flow simulator for complex geometries.

The core of my doctoral research was the development of technology for the simulation of ventricular cardiac electrophysiology. I published extensively in a wide range of fields: high performance computing, numerical analysis, software engineering, and cardiac electrophysiology. My contributions became the basis of multiple subsequent Ph.D. projects and were pivotal for the success of the EU- FP7 grant VPH-preDiCT. During that time, I was one of the lead developers of Chaste, an open source, multiscale, multi-physics simulation framework for systems biology. Chaste has been downloaded by academic and industrial research groups at over 600 locations worldwide.

Teaching: 
Effective Scientific and High Performance Computing with MATLAB (CoMPLEX M.Res. program):
 
The use of numerical computing environments like MATLAB is ubiquitous in science and engineering. MATLAB’s expressiveness and adequate level of abstraction makes it well suited for in silico modelling and simulation. In addition, MATLAB is often regarded as easier to learn by non-experts than traditional lower-level programming languages. Although in silico models are easy to develop in MATLAB, researchers often face difficulties when trying to scale them to higher dimensions or arbitrary geometries. This is not necessarily an intrinsic limitation of MATLAB, but often a consequence of an inadequate use of certain programming idioms or limited parallel capabilities.
 
At the end of this course, students will be capable of effectively scaling their in silico models to the level of performance required by small to medium scale simulation studies.
 
Syllabus and further details: here
Contact Email: 
miguel [dot] bernabeu [at] ucl [dot] ac [dot] uk

Publications