Vous êtes sur la page 1sur 53

Human Body Modeling with 

ANSYS Software

Marc Horner, Ph.D.
Technical Lead, Healthcare
ANSYS, Inc.
1 © 2011 ANSYS, Inc. October 24, 2011
Overview

This talk is motivated by the following observations:
1. Our understanding, and thus our ability to mathematically describe, the 
human body is to the point where one can assemble human body models as 
“boundary conditions” for  biomedical simulations.
2. Improvements in ease‐of‐use and stability of multiphysics modeling tools.
3. Computational capabilities are continually increasing.

2 © 2011 ANSYS, Inc. October 24, 2011
Types of Human Body Models
Human Body 
Modeling CARDIOVASCULAR MUSCULOSKELETAL DRUG DELIVERY EMAG

• CFD, FEA, and emag
models coupled to 
lumped parameter 
representations of the 
human body
Parametric System 
Level
• DOE for HBM
• Fully automated 
process Application areas: Application areas: Application areas: Application areas:
• Coronary stents • Orthopaedic implants • Transdermal • Medical imaging
• Peripheral stents • Exercise equipment • Inhalers • Cardiology
• Artificial organs • Oral • Drug delivery
• Intrathecal • Cancer treatments

3 © 2011 ANSYS, Inc. October 24, 2011
Types of Human Body Models
Human Body 
Modeling CARDIOVASCULAR MUSCULOSKELETAL DRUG DELIVERY EMAG

• CFD, FEA, and emag
models coupled to 
lumped parameter 
representations of the 
human body
Parametric System 
Level
• DOE for HBM
• Fully automated 
process Application areas: Application areas: Application areas: Application areas:
• Coronary stents • Orthopaedic implants • Transdermal • Medical imaging
• Peripheral stents • Exercise equipment • Inhalers • Cardiology
• Artificial organs • Oral • Drug delivery
• Intrathecal • Cancer treatments

4 © 2011 ANSYS, Inc. October 24, 2011
Idealized Blood Flow
Material properties
Can typically assume density and viscosity at average
hematocrit and high shear.
 = 1.05 g/cm3
 = 0.035 g/cm-s

Inflow conditions
Literature a good source for inflow data*

Resting Condition Exercising Condition


Qtot = 6.8 L/min, 75 bpm** Qtot = 12.8 L/min, 117 bpm**
Geometry * Stevens et al., Math Biosciences (2003)
5 © 2011 ANSYS, Inc. October 24, 2011 ** Murgo et al., Circ Res (2003)
Patient‐Specific Blood Flow
Material properties
Can take a blood sample and measure the patients
hematocrit, protein content, etc.

 vs H*  vs **

Inflow conditions
Taken from on-line measurements***

* Hinghofer-Szalkay, ?? (1986)
Geometry ** Cho & Kensey, Biorheology (1991)
6
(courtesy Materialise Inc.)
© 2011 ANSYS, Inc. October 24, 2011 *** Huntsman et al., Circulation (1983)
My Flow Patterns

speed pressure

wall shear
7 © 2011 ANSYS, Inc. October 24, 2011
Outflow Conditions For Aortic Flows
One complicating factor in aortic  Flow Rate Reqs for Human Organ Circuits1

flow modeling is the application 
of accurate outflow boundary 
conditions.
Specialized conditions are required 
+5.5 L/min
because the flow split at a 
bifurcation is controlled by  -0.3 L/min
downstream organ demand, not 
-0.8 L/min
the bifurcation geometry.
Without a specialized condition, an 
-1.1 L/min
analyst will not have a proper 
understanding of the baseline 
-1.1 L/min
flow patterns and how an 
implanted device affects those 
-2.2 L/min
patterns.

1Rhodes & Tanner, Med Physiology (1995)


8 © 2011 ANSYS, Inc. October 24, 2011
Modeling Options for Outflow BCs

Create a detailed geometric model of the 
entire cardiovascular system, extending 
from the heart to the capillaries.  The 
problems with this approach are obvious:
• Large geometric model required to 
capture full geometric details
• Accuracy of the geometry will be at 
question, esp. below 0.5 mm

9 © 2011 ANSYS, Inc. October 24, 2011
* Image from Texas Heart Institute web-site
Modeling Options for Outflow BCs
Electrical Circuit Model of the
Create a detailed geometric model of the  Human Cardiovascular System*
entire cardiovascular system, extending 
from the heart to the capillaries.  The 
problems with this approach are obvious:
• Large geometric model required to 
capture full geometric details
• Accuracy of the geometry will be at 
question, esp. below 0.5 mm
Use a lumped parameter approach to 
alleviate the need for detailed geometry 
at all length scales.
• The image to the right is an electrical 
circuit model of the human circulatory 
system, extending from the heart to 
the small arteries.
• Each box contains a lumped parameter 
representation of an artery section.

10 © 2011 ANSYS, Inc. October 24, 2011
* Westerhof et al., J Biomech (1969)
Lumped Parameter Model Details
The circuit on the lower left can be used to model the flow rate and 
pressure losses in each artery section. This circuit is referred to as 
a 3‐element Windkessel.

inertance term – accounts for


inertial effects (primarily in the resistance term – accounts for
large arteries) downstream flow resistance
(primarily in the capillary beds)

R RD
C

compliance term – accounts for


artery expansion/contraction
(primarily in the large arteries)
11 © 2011 ANSYS, Inc. October 24, 2011
The Westerhof Circuit in Simplorer

12 © 2011 ANSYS, Inc. October 24, 2011
Impedance Comparison

Westerhof Model Simplorer Model


(Inverted L Network)
Ansoft LLC total input imedance magnitude
500.00
Curve Info
mag(1/E1.I)
AC

400.00

300.00

mag(1/E1.I)
200.00

100.00

0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00
Ansoft LLC total input imedance phase
25.00 Curve Info
-ang_deg(E1.I)
AC

-ang_deg(E1.I) [deg]
0.00

-25.00

-50.00

-75.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00
F [kHz]

13 © 2011 ANSYS, Inc. October 24, 2011
Insert FLUENT into the Simplorer HBM

14 © 2011 ANSYS, Inc. October 24, 2011
Validation
‐ Flow Past a Symmetric Bifurcation
The Windkessel boundary condition was applied to an idealized representation of 
the abdominal aorta.  Transient values for the inlet flow rate and outlet 
pressures were culled from a literature source*.  The geometry and boundary 
conditions were symmetric in this case.  As seen in the figure on the right, there 
was excellent agreement between published and computational results for the 
outlet pressure.
Qin

Iliac RCR parameters

RD = 3.22 mm Hg-s/cc
R = 0.55 mm Hg-s/cc
C = 0.001 cc/mm Hg

Outlet pressure from SI[mplorerr compared


Pout well with the results of Taylor

* Wan et al. (preprint)


15 © 2011 ANSYS, Inc. October 24, 2011
A Non‐Symmetric Case

velocity

pressure
16 © 2011 ANSYS, Inc. October 24, 2011
Types of Human Body Models
Human Body 
Modeling CARDIOVASCULAR MUSCULOSKELETAL DRUG DELIVERY EMAG

• CFD, FEA, and emag
models coupled to 
lumped parameter 
representations of the 
human body
Parametric System 
Level
• DOE for HBM
• Fully automated 
process Application areas: Application areas: Application areas: Application areas:
• Coronary stents • Orthopaedic implants • Transdermal • Medical imaging
• Peripheral stents • Exercise equipment • Inhalers • Cardiology
• Artificial organs • Oral • Drug delivery
• Intrathecal • Cancer treatments

17 © 2011 ANSYS, Inc. October 24, 2011
Anybody for Implant Loads

AnyBody: Determine spine kinematics

ANSYS: FEM
analysis of a
spine implant
based on the
spine kinematics

AnyBody provides a detailed description of the loads and joint forces occurring in
the human body in daily activity. ANSYS Mechanical can import loads from Anybody,
providing critical information for testing orthopaedic implants and the like.
18 © 2011 ANSYS, Inc. October 24, 2011
Orthopaedic Workflow
Activities of daily living

Medical images

Optimization
Musculoskeletal
Simulation

CAD/Mesh Finite Element Analysis

FEA loads for


activities of daily living
19 © 2011 ANSYS, Inc. October 24, 2011
Functional Patient‐Specific Modeling

Activities of daily living

Boundary
Patient conditions
information

FE-model

20 © 2011 ANSYS, Inc. October 24, 2011
Unique Open Body Model Library

21 © 2011 ANSYS, Inc. October 24, 2011
Daily Activity Analysis

22 © 2011 ANSYS, Inc. October 24, 2011
Dynamic Physiologic Loads

23 © 2011 ANSYS, Inc. October 24, 2011
In vivo validation
‐ Hip forces

Thielen et al. 2009

24 © 2011 ANSYS, Inc. October 24, 2011
A Hip Simulator

25 © 2011 ANSYS, Inc. October 24, 2011
video from: http://edp.tkk.fi/en/research/tribology/research_projects/biotribology/
Slide Track Patterns
- Slide tracks represent the relative motion of the ball and cup

waveforms whole head patterns flattened patterns

Slide tracks on the head are


determined by mounting pens to
fixed points on the cup and vice
versa for slide tracks on the cup.

Upper fig: Calonius and Saikko, Acta Polytechnica (2003)


26 © 2011 ANSYS, Inc. October 24, 2011
Lower fig: Calonius and Saikko, J Biomech (2002)
Variation in Slide Track Patterns

27 © 2011 ANSYS, Inc. October 24, 2011
Model Inputs 
dhead = 20 mm
dcup = 22 mm
Geometry 
• Femur and implant provided in STL 
format
• Acetabular cup was created in ANSYS 
Mechanical

Initial STL geometry Implant and cup assembly

• Kinematic joint conditions provided by AnyBody:
140
23 Walking 120
Cycling
Angle in Degrees

13 HipAbduction 100

3 80 Flexion
HipFlexion
60 Abduction
-72.84 3.84 4.84
40 Ext. Rotation
HipExternalRot
-17 ation 20
0
-27
Time (S) -20 0 0.5 1

28 © 2011 ANSYS, Inc. October 24, 2011
Implementation
Geometry cup
• Head, cup (rigid), and stem
Material properties
stem
• Default material properties used for all parts head
Boundary Conditions
• Angular data entered as displacement BC’s for the 
cup in tabular format
Solver
• ANSYS Mechanical 12.1 transient solver
• Rigid‐flexible contact maintained between the head 
and the cup
Post‐processing
• Slide tracks are calculated using an APDL script which 
is inserted as a command object
29 © 2011 ANSYS, Inc. October 24, 2011
Creating Slide Tracks using APDL

Choose location of the “pen”
Get the node number for the pen 
location
Track the node during the transient 
cycle
Report (x,y,z) vs time for the node at the 
end of simulation
Use point locations to create keypoints
Join keypoints to form a spline
Display the spline with initial geometry 
to see the slide tracks

30 © 2011 ANSYS, Inc. October 24, 2011
Slide Tracks on the Cup
‐ Walking Profile

Side of cup Top of cup

31 © 2011 ANSYS, Inc. October 24, 2011
Slide Tracks on the Cup
‐ Cycling Profile

Side of cup Top of cup

32 © 2011 ANSYS, Inc. October 24, 2011
Slide Tracks on the Cup
‐ Walking Forces
Force vs time data
3500

3000 29
2500

2000

1500
50
1000
64
500 87
0 69
2 2.2 2.4 2.6 2.8 3 3.2
-500

-1000

-1500
MediolateralForce ProximoDistalForce
AnteroPosteriorForce

Slide track with time-step locations


33 © 2011 ANSYS, Inc. October 24, 2011
Slide Tracks on the Cup
‐ Walking Forces
Force vs time data
3500

3000 147
2500

2000

1500
168
1000 182
500 205
0 187
2 2.2 2.4 2.6 2.8 3 3.2
-500

-1000

-1500
MediolateralForce ProximoDistalForce
AnteroPosteriorForce

Slide track with time-step locations


34 © 2011 ANSYS, Inc. October 24, 2011
Slide Tracks on the Cup
‐ Kinematics

Side of cup Top of cup

35 © 2011 ANSYS, Inc. October 24, 2011
Wear Calculation

Archard’s Law defines the effect of sliding distance (v) and 
contact pressure () on the per cycle wear depth (w):
w( , )   k   ( , , t )  v( , , t )dt
cycle

Which can be approximated as:
w   kw    S
cycle
• where,
– kw is the wear coef. (which is material and surface dependent) = 1.066E‐
6
–  is the contact stress, and
– S is the sliding distance

36 © 2011 ANSYS, Inc. October 24, 2011
Cumulative Wear
(after 1 cycle)

walking cycling

wear depth reported in (mm)
37 © 2011 ANSYS, Inc. October 24, 2011
Validation
‐ Clinical Data
Hall et al (Proc Instn Mech Engrs, 1998) measured wear of 
UHMWPE acetabular components retrieved from 200 
patients.

38 © 2011 ANSYS, Inc. October 24, 2011
Types of Human Body Models
Human Body 
Modeling CARDIOVASCULAR MUSCULOSKELETAL DRUG DELIVERY EMAG

• CFD, FEA, and emag
models coupled to 
lumped parameter 
representations of the 
human body
Parametric System 
Level
• DOE for HBM
• Fully automated 
process Application areas: Application areas: Application areas: Application areas:
• Coronary stents • Orthopaedic implants • Transdermal • Medical imaging
• Peripheral stents • Exercise equipment • Inhalers • Cardiology
• Artificial organs • Oral • Drug delivery
• Intrathecal • Cancer treatments

39 © 2011 ANSYS, Inc. October 24, 2011
Pharmacokinetic (PK) Modeling

• PK models utilize a compartmentalized


approach to model drug distribution in the
body over time. The compartments
represent organs and other drug depots.

• PK models simulate drug/chemical:


• Absorption
• Distribution
• Metabolism
• Excretion

• Coupling the PK model to the drug


delivery model enables the designer to
further refine the delivery system to work
together with the human body, understand
dependence on patient variability, disease
state, etc.
40 © 2011 ANSYS, Inc. October 24, 2011
Transdermal Delivery Model
The system consists of a cylindrical patch applied to 
the skin.  The drug reservoir (in the patch) 
contains a mixture of drug and permeation 
enhancer.  Diffusion is the only mode of transport 
out of the patch and through the skin.  The far 
boundary of the skin is the inlet to the 
microcirculation, which acts as a drug sink.

• Two boundary conditions are tested 
at the far wall:
1. zero flux condition – which  patch
assumes the plasma is an infinite  skin
sink
2. PK condition – to account for drug  microcirculation
build‐up in plasma

41 © 2011 ANSYS, Inc. October 24, 2011
Transdermal Model Details
Axisymmetry is assumed
r = 0.9 cm
Drug and enhancer are modeled as  user‐defined 
scalars in FLUENT patch 50.8 m

A transient diffusion equation models transport  skin 50.8 m


through the skin and patch:
r = 0.925 cm
ci
   Di ci 
t
, where i = drug, permeation enhancer

• The interfacial conditions at the patch‐skin interface are continuity of flux 
and a partitioning (jump) condition:
Di , patch  ci , patch  Di , skin  ci , skin
c drug , skin / c drug , patch  K , where K is the partition coefficient

42 © 2011 ANSYS, Inc. October 24, 2011
Transdermal Model Details
‐ Formulation Effects
The permeation enhancer increases the drug flux through the 
skin.  The level of enhancement is typically a function of the 
local enhancer concentration.  Two typical situations are:
Ddrug,skin = Ddrug,0+*Cpe Kdrug,skin = Kdrug,0+*Cpe

 increasing
 increasing

The permeation enhancer increases The permeation enhancer increases


the diffusivity of drug in the skin drug solubility in the skin

43 © 2011 ANSYS, Inc. October 24, 2011
Validation Case
‐ Fentanyl Patch, No PK Model
Rim et al. examined the effect of enhancement type on drug 
flux.  These plots compare FLUENT results to their 
experimental and computational results.
Comparison of drug  flux through  lower boundary  of epidermis.  Initial  Comparison of drug  flux through  lower boundary  of epidermis.  Initial 
concentration of drug and  enhancer  0.06, 0.04(gm  cm‐3). dt=100,  concentration of drug and  enhancer  0.06, 0.04(gm  cm‐3). dt=100, 
ncells=58400, µ=1.8e‐4, η=0.0 ncells=58400, µ=0.0, η=19.0
2.5 2.5
Drug Flux (µg cm‐2 hr‐1)

Drug Flux (µg cm‐2 hr‐1)


2.0 2.0

1.5 1.5

1.0 1.0
simulation simulation

0.5 experiment 0.5 experiment


reference reference
0.0 0.0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (hr) Time (hr)

flux when Ddrug=f(Cenhancer) flux when K = f(Cdrug)

Reference: Rim et al., Annals of BME, (2005)


44 © 2011 ANSYS, Inc. October 24, 2011
Pharmacokinetic (PK) Analysis of a 
Fentanyl Patch
• Patch model extensions:
MOTIVATION: Understand/optimize
patch performance by including the • BC between epidermis and
dermal vasculature:
physiologic processing of drug.
c 
D ︵
 P  [c l , t)  fu  Cp (t)]
z
CFD Model

• The ODE’s for the three 
compartment PK model are:

plasma: dC ︵
p dt    [ c l , t )  f u  C p ( t )]

 [ k el  k12  k13 ]  C p ( t )
 k 21  2C 2 ( t )  k 31  3C3 ( t )
PK Model

dC 2 dt  k12C p ( t ) /  2  k 21C2 ( t )
well-perfused
compartment:

dC 3 dt  k13C p ( t ) /  3  k 31C3 ( t )
poorly perfused
compartment:

45 © 2011 ANSYS, Inc. October 24, 2011
Pharmacokinetic (PK) Analysis of a 
Fentanyl Patch
• Patch model extensions:
MOTIVATION: Understand/optimize
patch performance by including the • BC between epidermis and
dermal vasculature:
physiologic processing of drug.
c 
D ︵
 P  [c l , t)  fu  Cp (t)]
z
CFD Model

• The ODE’s for the three 
compartment PK model are:

plasma: dC ︵
p dt    [ c l , t )  f u  C p ( t )]

 [ k el  k12  k13 ]  C p ( t )
 k 21  2C 2 ( t )  k 31  3C3 ( t )
PK Model

dC 2 dt  k12C p ( t ) /  2  k 21C2 ( t )
well-perfused
compartment:

dC 3 dt  k13C p ( t ) /  3  k 31C3 ( t )
poorly perfused
compartment:

46 © 2011 ANSYS, Inc. October 24, 2011
Types of Human Body Models
Human Body 
Modeling CARDIOVASCULAR MUSCULOSKELETAL DRUG DELIVERY EMAG

• CFD, FEA, and emag
models coupled to 
lumped parameter 
representations of the 
human body
Parametric System 
Level
• DOE for HBM
• Fully automated 
process Application areas: Application areas: Application areas: Application areas:
• Coronary stents • Orthopaedic implants • Transdermal • Medical imaging
• Peripheral stents • Exercise equipment • Inhalers • Cardiology
• Artificial organs • Oral • Drug delivery
• Intrathecal • Cancer treatments

47 © 2011 ANSYS, Inc. October 24, 2011
HFSS‐Workbench Link for Modeling 
Cancer Treatment
Hyperthermia cancer treatment used to accelerate effects of 
chemotherapy
RF power applied to tumor using phased array antenna
• Eight strip dipole antennas connected in parallel pairs and printed on 
inner surface of cylindrical plastic shell

HFSS Model of Patient with Phased Array MRI Cross Section of Lower Leg with Tumor
Applicator on Lower Leg (Courtesy Duke University Medical Center)

48 © 2011 ANSYS, Inc. October 24, 2011
HFSS Model of RF Phased Array 
Applicator
Device operates at 138 MHz
• Element excitations optimized to concentrate power inside tumor 

Electric Field Magnitude with


Optimized Array Weights

HFSS Model of Applicator on Leg


HFSS Model of RF Phased Array Tumor (shown in green)
Applicator

Local Specific Absorption Rate (SAR)


49 © 2011 ANSYS, Inc. October 24, 2011 with Optimized Array Weights
Transient Thermal Analysis in ANSYS 
Workbench

Results of Transient Thermal Analysis


Geometry in ANSYS Workbench (max/min temperature in the tumor)

• Electromagnetic power is a function of 
time
– 28 W used to reach desired
temperature and 9 W to maintain
temperature
• Transient analysis used to determine 
temperature rise inside tumor
– Reaches 47° C in 6 minutes and is
maintained there for 14 minutes
Results of Transient Thermal Analysis
50 © 2011 ANSYS, Inc. October 24, 2011
Closing Remarks

Coupling detailed simulations to 
lumped parameter models can 
provide more rigorous prediction of 
device performance.

51 © 2011 ANSYS, Inc. October 24, 2011
Closing Remarks

Coupling detailed simulations to 
lumped parameter models can 
provide more rigorous prediction of 
device performance.
V&V is possible.

,µ ,η
2.5

Pout
Drug Flux (µg cm‐2 hr‐1)

2.0

1.5

1.0
simulation

0.5 experiment
reference
0.0
0 10 20 30 40 50 60
Time (hr)
52 © 2011 ANSYS, Inc. October 24, 2011
Closing Remarks

ANSYS provides flexible and open solutions that 
can be adapted to solve even the most 
challenging problems facing the biomedical 
engineers of today.
The ANSYS vision for this industry includes 
collaborating with other software vendors 
to enable one‐way and/or cosimulation
with the highest fidelity models available.

53 © 2011 ANSYS, Inc. October 24, 2011