Modelling cardiovascular physiology Fantastic voyage (1966) Dr Emma Chung (emlc1@le.ac.uk) University of Leicester Department of Cardiovascular Sciences Distance dramatically slows diffusion The rate of diffusive transport is important, because nutrient delivery needs to keep up with metabolic demand Einstein’s theory of Brownian Motion (1905) tx Example Distance (x) Time (t) Capillary wall 1.0 m 510-6 s Capillary to cell 10 m 0.05 s Skin to artery wall 1 mm 9.26 min Left ventricular wall 1 cm 15.4 h Table showing diffusion times for glucose. (NB. for glucose in water at 37C D=0.9 10-5, and for oxygen in water D=310-5 cm2s-1) 2 2D D is the solute coefficient, x is distance and t is time Most cells lie between 10 and 20 m of the nearest capillary Evolutionary biology Very simple organisms like the hydra exchange oxygen and nutrients purely by diffusion In larger organisms the organs are supplied via a ‘closed cardiovascular system’ with varying numbers of hearts! Many molluscs and invertebrates have an ‘open circulatory system’ (e.g. Snails) 3 Vasculogenesis in the embryo The initial ‘setting up’ of the blood vessels is dictated by the genes. At day 21 of embryogenesis, the heart begins beating and vascular remodelling starts to occur. Pressure, flow patterns, shear stress and metabolic demand then interact with genetic and environmental factors to produce the mature vasculature. Optimising the delivery of fluid by branching 5 Branching angle (Murray’s law 1926) Optimal branching angle depends on fluid dynamical efficiency at the bifurcation. Branches with smaller diameters make larger branching angles Murray's law (1926) The hydraulic conductance per blood volume of the cardiovascular system is maximized when the cube of the radius of a parent vessel equals the sum of the cubes of the radii of the daughters. ‘Small side branch’ Coronary arteries • Genetic code • Physiology • Fluid dynamical efficiency AIM: Simulate optimised 3D vascular trees for any arbitrary tissue or organ For inclusion in simulations of blood flow and autoregulation Medical applications include the design of 3D printed artificial organs (e.g. for kidney replacement) 3D bio-printing of replacement organs Making a bit of me: A machine that prints organs is coming to market The Economist, Feb 18th 2010 “the company expects that within five years, once clinical trials are complete, the printers will produce blood vessels for use as grafts in bypass surgery. With more research it should be possible to produce bigger, more complex body parts. Because the machines have the ability to make branched tubes, the technology could, for example, be used to create the networks of blood vessels needed to sustain larger printed organs, like kidneys, livers and hearts.” - Keith Murphy, Organovo's Chief Executive Mini-human brains only 4 mm wide grown from stem cells “The mini-brains can not grow larger than a few millimetres because of a lack of blood supply” Lancaster et al. Cerebral organoids model human brain development and microcephaly. Nature 501, 373–379, 2013 Constrained Constructive Optimisation (CCO) The current standard technique for generating arterial trees is called constrained constructive optimisation In silico vascular tree grown using constrained constructive optimisation (optimised using genetic algorithms) Jonathan Keelan, Jim Hague and Emma Chung Coronary arteries • Anatomy of the major coronary arteries is similar between individuals • Large trunk vessels asymmetric branching • Only fine vessels venture into the tissue Modelling the coronary vasculature - CCO CCO growth of the first few layers of coronary arteries to supply a ‘myocardial substrate’ simulating the left atrium Jonathan Keelan, Jim Hague and Emma Chung • • • Too symmetrical Too many ‘constraints’ required (especially for hollow organs) No guarantee that the final solution is globally optimised Simulated annealing (SA) Simulated Annealing (SA) techniques can be used to obtain a good approximation to the global optimum of a given function in a large search space. Modelling the coronary vasculature – Simulated Annealing Coronary arterial trees grown using Simulated Annealing to achieve global optimisation Only need to define: • Geometry and metabolic demand of the tissue substrate • Position of the start nodes • Number of nodes Looks much better, but how can we make a quantitative comparison with real coronary arteries? • Vessels above a certain size were not allowed to penetrate the tissue Strahler (stream) Order Within the Strahler Order scheme the lowest Strahler order numbers correspond to the smallest arterioles and the largest numbers to major vessels. Vessels of the same Strahler Order in the simulated coronary arterial trees were grouped. The average length, diameter and branching properties were then compared to existing morphological data from Kassab et al. 1993 Results of comparison with pig coronary arteries Vessel diameter Vessel length Artefact from early termination Simulated annealing provided an almost perfect match to pig morphological data for vessel lengths and diameters. Kassab et al. Morphometry of pig coronary arterial trees. American Journal of Physiology, 265(1):H350{65, 1993 Results of comparison with pig coronary arteries Ratio of diameter of smallest daughter (DS) to parent vessel (DP). For the major arteries there tends to be a major trunk vessel of similar size to the parent vessel with a small side branch. As vessels become smaller (i.e. For lower Strahler orders) the branches become more symmetric. Modelling the cerebral arteries “The brain is the most complex structure in the universe” • Very high density of vessels Grey matter makes up less than 1% of the mass of the body but accounts for nearly 20% of total oxygen consumption • Two types of tissue substrate (grey and white matter) • Fluctuating metabolic demand • Complex geometry with boundaries • Position of the start nodes?? Anatomy varies between individuals ACAs MCA BA Posterior communicating arteries VA Relative sizes of the vessels There is no typical ‘textbook’ circle of Willis Variations in the circle of Willis identified by Kleiss (1941). Wax models and static phantoms of the circulation 18th century Wax model of the arteries from La Specola museum, Florence, Polymer cast of the cerebral circulation from Gunter von Hagens’ “Body Worlds” exhibition. Scanning electron microscopy of a corrosion cast of the cerebral microcirculation. (La Torre 1998) With modern imaging methods we can create patient specific models Vascular phantom fabrication Rapid prototyping CAD package Virtual patient simulation Imaging data Laboratory vascular replica Vascular phantom • Programmable pump unit • Injection ports • Silicon model of the major arteries • Distal resistances controlled by length and diameter of outlet tubing • Reservoir containing a Blood Mimicking Fluid (BMF) Electrical circuit analogue approaches V=IR Electrical circuit analogue ‘Idra’ ‘Electra’ Study of cerebral hemodynamics by analogic models Fasano et al. (1966) Electrical circuit models R 8l a 4 Perfusion territories supplied by the circle of Willis Arterial Spin Label imaging with MRI Hendrikse et al. Cerebral autoregulation Bayesian analysis of cerebral blood flow autoregulatory mechanisms for personalised medicine. Computational fluid dynamics Computational Fluid Dynamics (CFD) simulations of blood flow require a well-defined mesh to model anatomically realistic vasculature,and are good at modelling complex haemodynamics. BUT... they struggle to model solid-fluid interactions: • • • • • Courtesy of Quang Long (Brunel) Interactions between bloodflow and vessel wall motion Cardiac valve motion Thrombus and plaque formation Particulate nature of the blood Motion of objects in the flow Do emboli ‘go with the flow’? Results: 25% 1000 m emboli: 98:2 500 m emboli: 93:7 200 m emboli: 89:11 MCA:ACA flow: 75:25 75% http://news.bbc.co.uk/1/hi/health/8516802.stm High-speed video recorded at 300 fps and slowed to 30 fps (1 cardiac cycle every 10 s) Chung et al. Preferred embolus trajectory through a replica of the major cerebral arteries Stroke 2010;41:647-652 Preferred embolus trajectory 75 30.2 2 8.4 Watershed regions 0 3.4 10% 20% 5 18 30% 18 40 40% Emboli 100 100 Flow Emboli are more likely to come to rest at the tips of the major arteries and branches Brain injury after cardiac surgery Patient 13 Patient 19 Patient 63 Before surgery 6-8 weeks after surgery Subtraction image Registration and subtraction of ‘before’ and ‘after’ 3-T MRI FLAIR images to distinguish between ‘old’ and ‘new’ lesions Brain injury after cardiac surgery New lesions following surgery are relatively minor compared to the number and volume of pre-existing lesions due to chronic cardiovascular disease. Brain injury after cardiac surgery Composite image of new lesions observed in 24 patients after cardiac surgery Smoothed-particle hydrodynamics (or: How we learned to stop worrying (about meshes) and love particles) Computational fluid dynamics involves building a mesh and watching how the fluid flows past the mesh. Smoothed Particle Hydrodynamics (SPH) is a mesh-free Lagrangian method (where the coordinates move with the fluid) Smoothed-particle hydrodynamics Tom Hands, 4th year Physics project Monte-Carlo simulation of bubbles moving through the cerebral vasculature Capillary mesh ~ 6 m 12 m n=3 n=5 n=1 β = 2.3 Major arteries > 1 mm 1 mm History of flow in end arterioles. Fully blocked clusters are shown in black and free flow in white. Featured article: Hague JP, Banahan C and Chung EML. Modelling of impaired cerebral blood flow due to gaseous emboli. Phys. Med. Biol. 2013 58:4381-4394 Monte-Carlo simulation of bubbles moving through the cerebral vasculature Timing and diameters of bubbles entering the middle cerebral arteries in a 55 year old patient during combined atrial valve replacement and coronary artery bypass graft surgery. Monte-Carlo simulations are used to estimate the potential impact of bubbles on MCA blood flow. Funders: Collaborators: • • • • Jim Hague (Open University) Mark Horsfield (Leicester) Kumar Ramnarine (UHL) Ronney Panerai (Leicester) • Engineering and Physical Sciences Research Council (EPSRC) • British Heart Foundation • Wellcome Trust Post-docs: • NIHR Leicester Cardiovascular Biomedical Research Unit • • • TH Wathes Foundation • Nihon Kohden (Japan) Caroline Banahan (UHL) Baris Kanber (Leicester) PhD students: • • • Nikil Patel (Leicester) David Marshall (Leicester) Jonathan Keelan (Open University) Undergraduate students: • • Tom Hands Katie Masters Thank you!