Post on 16-Jan-2016
Biomechanics and biology: bridging the
gap
Sam EvansSchool of Engineering
EvansSL6@cardiff.ac.uk
Bone strength and GH
•How do changes in growth hormone and other endocrine factors affect bone strength?
•How to test the strength of the bones?
Problems
•Many groups of bones to test
•Considerable variability
•Small, weak bones
How to do it?•Impractical to machine tensile test specimens from bones
•Need to test whole bones
•3 point bend tests simulate a typical fracture scenario
•Need to measure cross- section to determine geometric contribution to strength (2nd moment of area)
Second moment of areaIrregular shape-
elliptical assumption gives a reasonable approximation for most bones
Examples• Femoral
bending strength
• Squares: male• Circles: female• White:WT• Black: TGR
Evans et al, Journal of Bone and Mineral Research 18(7) (2003): 1308-16.
Challenges
Understandable results
Geometric and material variability
Many variables and specimens
?
Engineering rigour
Future challenges?
•High throughput automated testing•Specimen- specific measurements
and models using imaging•Fast, reliable computer models
needed to analyse the results•Better standards for measurement
and modelling
Conclusions
•Working with biologists is fun!
•Many opportunities for interesting research
•Possibilities for big advances using new technology
Examples
Effect of growth hormone deficiency on bone strength
Body weight (grams)
400380360340320300280260240
Yie
ld lo
ad (
N)
140
120
100
80
60
40
GROUPTgRRsq = 0.0943
ControlRsq = 0.5367
Variability
•Need detailed, subject specific measurements and models for accurate results
•Many tests needed to achieve statistical significance
•Sophisticated material models required
Many variables and specimens
•Many groups often need to be tested
•Many biological questions
•Complex problems require many tests to investigate interaction of multiple variables
•High throughput testing is needed
Understandable results
• Results need to be understood by biologists
• Even expert engineers struggle with complex models of soft tissues etc
• Requires models that are no more complex than necessary
• Good explanations needed too
Methods•Standard testing machine with 100N load cell
•A range of 3pb fixtures in various sizes
•Eight measurements of cross section using travelling microscope to determine second moment of area
Stress and strain
0
20
40
60
80
100
1 1.2 1.4 1.6 1.8 2
Engineering (Biot) stress
True (Cauchy) stress
What do we want to do?
•Often need to predict or measure the mechanical behaviour of biological materials
•eg implant design, development of surgical procedures
•Measuring effects of biological changes
What do we need?
Constitutive model
=E =-T/L
Stress analysis
=F/A =L/L
Behaviour of material
Measurements
So what’s the problem?•Standard tests assume a
simple stress analysis which in turn assumes a simple constitutive model
•Not valid for inhomogenous, anisotropic materials!
•More tests needed, and more complex analysis
Any other problems?•Most biological materials change
irretrievably in vitro•Cutting specimens disrupts their
structure and alters their behaviour
•Human tissue often different from animals
•Need clinical measurements• Ideally want to test in vivo
What do we need?
Complex constitutive
model, many parameters
Numerical simulation
Behaviour of material
3D, time dependent
measurements
Testing and simulation
• Can’t simulate tissue behaviour without measuring parameters
• Can’t test tissue without some sort of model or simulation
• Need to validate simulations• Testing & simulation are linked• Has testing been neglected?
Cells with various aspect ratios.
Load more or less independent of length
Fatigue of bone cement
Disc replacement
Periodontal ligament• In vivo testing of human
periodontal ligament (PDL) using small scale motion analysis
• Development of sophisticated computational models at UWCM
• Funded by EPSRC
-4
-2
0
2
4
6
8
0 1 2 3 4 5
Actual
Measured (m)
-50
0
50
100
150
200
250
0 50 100 150 200
Measured (m)
Actual
Conclusions
•Simulation and testing must go hand in hand
•Many standard methods assume a model that may be invalid
•We need to work together!