Download - Computational Biomechanics

8/11/2019 Computational Biomechanics

1/1010 BIOMEDICALCOMPUTATION REVIEW Winter 2006/07 www.biomedicalcomputationreview.org

ComputationalBiomechanicsComputationalBiomechanics

Making StridesToward Patient Care

BY REGINA NUZZO, PhD

Simulations of gait, like the one illus-

trated here, are being used by

researchers in biomechanics to quanti-

fy how individual muscles contribute to

an observed movement. This simula-

tion represents one of the most

detailed muscle-actuated simulations

generated to date. The motions of the

body segments are generated by

forces applied over time by about one

hundred separate muscles. The

insights gained from simulations

like this one are helping shed new

light on movement abnormalities

commonly observed in individu-als with cerebral palsy, stroke,

and other neuromuscular

impairments. Image courtesy

of Chand John, Frank C.

Anderson, and Scott Delp.


2/10Winter 2006/07 BIOMEDICALCOMPUTATION REVIEW 11www.biomedicalcomputationreview.org

To understand how muscles contract

and joints flex, researchers have dis-sected cadavers and experimented

with animals. They can describe howbones, muscles, and tendons connect in acomplicated geometry; how mus-cles exert forces on joints; andeven how sparks in the brain cantrigger a muscles contraction.

Meanwhile, and mostly inde-pendently, clinicians have beentreating people for sportsinjuries, stroke, and movementdiseases such as cerebral palsy

and osteoarthritis. Using trialand error, theyve assessed whichrehabilitative strategies and sur-gical interventions work best.

Until recently, these two per-spectives have not been well inte-grated. Clinical observationsmight miss the interplay of forcesthat lead to an injured knee,

while static equations regardingthe flexion of a dead mans kneemay by themselves be of littlehelp in treating a torn ligament.Researchers were missing the cause-and-

effect models that linked the physicalforces with the clinical outcomes.

Now, computational models are bridg-ing the gap between micro and macro,between brain signals and gross humanmovement, and most recently, betweenresearch and clinichelping to bring bio-mechanics knowledge from the bench allthe way to the bedside. Computationalbiomechanics moves us beyond intuition,so that were able to understand how abasic cadaver study might relate to amovement disorder, says Frank (Clay)

Anderson, PhD, senior research associ-ate at the Neuromuscular BiomechanicsLab at Stanford University.

Computer models can let surgeonsexplore different types of surgery on apatient before attempting the realthing. Or generate graphic simulationsof athletes sprinting or cerebral palsypatients walking with a stiff-kneed gait.Some models can provide personalizedrecommendations for rehabilitation;others let researchers design new and

improved implants that last longer than

ever before. Flexible models allowresearchers to tweak variables and askwhy? and what if? in ways that bringthe power of computational biome-

chanics straight to the patientpickingup where intuition and trial-and-error

treatments leave off.

BUILDING ASHAREDFOUNDATION

Unlike their forebears in the field oftraditional biomechanics, many computa-tional biomechanics researchers comefrom an engineering background. Wheninnovative engineers tried to apply stan-dard mathematical and computationaltechniques to interesting biological prob-lems, they soon learned that what worksfor planes and cars doesnt always work

for muscles and bones. So the first chal-lenge for these new computational bio-mechanists was to learn the language ofbiology. Starting in the 1970s, they duginto biology and began sharing theirtoolsmultibody dynamics, control theo-ry, and finite element modeling, for exam-plein journals and at conferences thathad previously served only physical thera-pists, rehabilitation medicine experts,neurologists, and orthopedic surgeons. Bythe late 1980s, the rehabilitation commu-

When innovative engineers

tried to apply standard

mathematical and

computational techniques

to interesting biological

problems, they soon learned

that what works for planes

and cars doesnt always work

for muscles and bones.

Walk, run, bend, reach.The elements of human movement have

fascinated research scientists for centuries.



nity had welcomed these new approaches,and computational biomechanics as afield had come into its own.

At the same time, advances in com-puting were infusing the field with new-

found power. With the increases in com-puter speed in the late 1980s and 1990s,researchers could suddenly analyze theirbiomechanics models computationally

with numerical algorithms rather thanjust analytically with pencil-and-paper

mathematics. In the mid-1990s, parallelcomputing took off, and within thedecade new and better algorithms, bor-rowed from control theory in robotics,

were available. Ten years ago, a pair ofsimulated legs walking a single step wouldtake approximately 1,000 hours to com-pute. Today, researchers can churn outsimilar calculations in about an hour.

But even from the beginning, oneimportant challenge faced computation-al biomechanics researchers: the need todevelop ways to work with each other.

With no standard modeling or softwareframework in the community, computa-tional biomechanics labs were spendingup to five years developing a single mus-culoskeletal model. Collaborating andbuilding off of others progress was near-

ly impossible. Everyone had to reinventthe wheel, one graduate student at atime, says Scott Delp, PhD, professorand chair of bioengineering and head ofthe Neuromuscular Biomechanics Lab

at Stanford University. It really held upthe field.

To help remedy that situation, in theearly 1990s, Delp and Peter Loan, a bio-

mechanist and software engineer, intro-duced a new modeling environment,

called Software for InteractiveMusculoskeletal Modeling (SIMM). Thisgeneral software package let users create,alter, and evaluate models of many struc-turesall within a common platform.Since then, hundreds of labs have adopt-ed SIMM to create and analyze comput-er models that range from a humanclimbing stairs to a Tyrannosaur nimblytrotting on hind limbs. The commontool also allowed researchers to moreeasily share models, and interdiscipli-nary studies started to focus on clinical

applications of biomechanics.At the same time, funding agencieswere taking notice. Biomedical comput-ing toolsand the development of aninfrastructure to support thembegan tofind a deserving place alongside tradi-

tional laboratory science. In 2004, NIHannounced the award of $79.7 million infunding to establish four NationalCenters for Biomedical ComputingOne of these was Simbios, the Center for

Physics-Based Simulation of BiologicalStructures at Stanford University, whichincludes neuromuscular dynamics as adriving biological project.

The neuromuscular dynamics groupat Simbios, in an effort led by Delp and

Anderson, is now developing OpenSiman open-source software system for biomechanical modeling and simulationFreely available and incorporating mod-ern plug-in technology, the new softwarecould be a boon for researchers who needsimulation software. Having a commonplatform will radically change the field,says Thomas Buchanan, PhD, a professor of mechanical engineering atUniversity of Delaware. This will reallytake the field in a whole new direction.

REPLACING AND

REPAIRINGJOINTSThe design of joint replacements is

one of the most mature applications ofcomputational biomechanics. Ideally, artificial hips and knees need to withstand

Computational biomechanics moves us beyond intuition, so that

were able to understand how a basic cadaver study might relate

to a movement disorder, says Frank Clay Anderson.



in a diverse set of situationsin patientsranging from frail to heavy-set, for activitiesfrom sitting down to climbing stairs.

Bartel and his team build computa-tional models that incorporate vari-

ables surgeons and implant designerscan manipulate, such as implant mate-rials and geometric design, as well asenvironmental variables beyond their

control, such as a patients varyingactivity level and individual bonestrength. Collaboration with ThomasSantner, PhD, a professor of statistics

loads up to several times a persons bodyweight and should function seamlesslyover a lifetime. Forty years ago, however,joint replacements were far from perfect.Implants often wore down or broke out-

right under the rigors of everyday use.Then, with the help of advanced mus-

culoskeletal computer models, researchersstarted to better understand the loadsapplied to natural joints undernormal conditions. When theybrought these models to thedrafting table, they were able todesign better artificial joints. Bythe 1980s, prosthetic joints hadimproved dramatically. Now, arti-ficial joints far outperform theirold-style predecessors in durabili-

ty and reliability.Today, researchers continue to

improve implant designs inincreasingly refined ways. DonaldBartel, PhD, a professor ofmechanical and aerospace engi-neering at Cornell University, is

working to build real-world variability intothe prosthetic design process. He uses bio-mechanical and statistical models to createartificial hips that are designed to function

at Ohio State University, has led torobust statistical optimization methodsthat consider both the design variablesand the environmental variables andthen optimize the design of the

implant. The methodology allows us

In 1995 it took 4,000 hours of parallel computing time to generate a half a second of walk-

ing in the image shown here. Contrast that with the transparent walker shown on themagazine cover and the previous pages: With twice the number of muscles, 2 seconds of

walking took 20 minutes on an ordinary desktop PC. Advances in computing speed pro-

duced some of these improvements, but mostly they result from advances in algorithms.

Images, courtesy of Frank (Clay) Anderson, come from Andersons dissertation work at the

University of Texas, Austin, under Marcus G. Pandy.

Having a common

platform will radically

change the field, says

Thomas Buchanan.

This will really take

the field in a whole

new direction.



to say whether variation in the environ-mental variables will swamp out any-thing we do in the design, Bartel says.It helps us decide where to put ourefforts and resources.

In addition to designers, orthopedicsurgeons may benefit from computation-al biomechanics. A common surgery forhip dysplasia involves cutting out a por-tion of the hip socket and reorienting itto provide better alignment with thefemoral head. The hope is that the proce-dure will distribute stresses in the hipmore evenly, thus relieving the patientspain and also decreasing the wear andtear on the hips cartilage. But the surgeryis tricky and not always successful.Surgeons have been guided mostly bytheir own experiences and intuition.Theres little attention paid to the actualmechanics of the hip, either in its initialstate or how it might be altered after sur-

that the standard surgical evaluationprocess for hip dysplasia patients willinclude a model of the patients individual stress-strain distribution, which pre-dicts how the joint will change underincreasing loads. That detailed informa-tion could then steer surgeons in deciding not only which patients are likely tobenefit from surgery, but how to plan thesurgery itself. Weiss and his team have

already started to put this into practiceSo far, theyve collected CT scans from 15hip dysplasia patients as well as twohealthy volunteers and are using theimages to build models of their hips.

PHYSICALTHERAPYAPPLICATIONS

Sometimes joint injuries or otherjoint problems are better resolved without surgical intervention. For example, acommon, painful knee soft tissue

gery, saysJeffrey Weiss, PhD, associateprofessor of bioengineering and orthope-dics at the University of Utah.

To help surgeons make better deci-sions about where to cut, Weiss and hiscollaborators at University of UtahChris Peters, MD, professor of ortho-pedics, and Andrew Anderson, doctor-al student in bioengineeringtailor com-putational biomechanical models toindividual patients. They use CT imagesand biomechanical measurements todetermine a patients specific hip geom-etry, standing loading patterns, andjoint reaction forces. Using finite ele-ment analysisa common way to modelthe mechanics of systems with complexgeometriesthe model predicts the con-tact stresses at the hip during inactivityas well as during everyday activities.

Ultimately, this process might becomeroutine. Weiss and his colleagues hope

Finite element predictions of bone displacement in the direction of primary joint loading between a normal pelvis (left) and a pelvis with

acetabular dysplasia (right) during simulated walking. The pelvic bone of the patient with hip dysplasia deformed much more than that of

the normal hip joint, indicating that hip dysplasia may predispose the joint to early degeneration. Courtesy Jeffrey Weiss.



injurya torn anterior cruciate ligament(ACL)often requires invasive recon-structive surgery. But, mysteriously, someathletes can recover and perform well

without it. To find out how copers

manage to compensate for their damagedACL, Buchanan and his collaboratorsare combining musculoskeletal models

with patient data from MRI scans, elec-tromyographic (EMG) recordings, andkinematic experiments. Our hypothesisis that some people have a way of usingtheir muscles that allows them to stabi-lize the knee joint without relying onthat ligament, Buchanan says. Thequestion is whether its possible to trainother people to use the same technique.

Indeed, biomechanics simulations

might be a boon for individualizedphysical therapy. B.J. Fregly, PhD, anassociate professor of mechanical andaerospace engineering at the Universityof Florida, is interested in designingcustomized treatments for patients with

knee osteoarthritis. In fact, Fregly him-self suffers from early-stage osteoarthri-tis, a remnant from his days of soccer,basketball and track in high school andcollege. Treatment for osteoarthritis

often involves replacing the damagedknee with an artificial joint oronlyslightly less invasiveundergoing a hightibial osteotomy. The latter is a proce-dure in which a surgeon cuts the tibiaclose to the knee and adds or removes a

wedge of bone, so that the patient thenbecomes slightly knock-kneed. The goalis to shift some of the load away fromthe diseased inner portion of the kneeand place it on the still-healthy outerportions of the knee. Fregly originally

wanted to use computational biome-

chanics to help surgeons better plan forhigh tibial osteotomies in individualpatients. But when he ran up against adearth of surgical patients to study,Fregly decided to ask a different ques-tion: Could he plan an individualized

therapy program that would give himthe same benefits as high tibial osteoto-my surgery?

First, Fregly andJeff Reinbolt, PhD,now a postdoc at Stanford, built a com-

puter model of Freglys gait. They tooka general, full-body walking model andcalibrated it to his own movement data.

After an iterative process of adjustingthe model to match each joint individu-ally and then fine-tuning the entiremodel globally, Fregly had a walkingsimulation that closely resembled hisown natural stride. Then the team usedinverse dynamic optimization, a numer-ical technique often used in computeranimation, to predict the overall jointloads and motions that would minimize

the load on the inside of the knee. Theyused this optimizer to determine whichsmall changes in Freglys gait would like-ly reduce the loading in his knees inmuch the same way that a high tibialosteotomy would.

To reduce damaging loads on the inner side of his own early-stage osteoarthritic knee, B.J. Fregly measured the moment arm of the ground

reaction force vector relative to his knee center before training (above left); then predicted the optimal moment arm (above center); and

then trained himself to walk in a way that mimicked the optimum (above right). Note that the distance between the left knee center and

the ground reaction force vector (in green) is reduced (relative to pre-training) by the optimization. This reduction is comparable to what

the patient (Fregly) achieved post-training. The moment arm reduction for both the optimization and post-training resulted in roughly a

40 percent reduction in the first peak of the knee adduction moment curve, which is likely to be clinically significant. Courtesy of B.J. Fregly.



tial mechanisms of injury during athlet-ic activities. Thelen and BryanHeiderscheit, PT, PhD, assistant pro-fessor in the department of orthopedicsand rehabilitation at University of

Wisconsin, recently investigated ham-string strain injuries in sprinters.

They first used motion captureequipment to measure whole-body

motion while athletes sprinted on ahigh-speed treadmill at speeds up to 20mph. They then fit scaled muscu-loskeletal models to the measuredmotion, which allowed them to estimatehow the hamstring muscles arestretched during sprinting. They haveshown that the greatest hamstringstretchand potentially the time ofgreatest risk for injuryoccurs late in

the swing phase of the sprintinggait cycle. This is when the foot

is still off the ground and theleg is being decelerated by thehamstrings before hitting the

ground again.To better understand

the dynamics of themovement, Thelen usedan algorithm developed

with Anderson to effi-ciently compute the mus-

cle excitation patterns thatproduce the measured hip and

knee motions of their sprint-ing volunteers. The muscleexcitation patterns were

inputs to a simulation ofthe crucial late sprinting

stage. The model, likemany other simu-

lations of human movementused a standard, computationally intensive technique calledforward dynamics, in whichthe differential equations of a

dynamic model are integratedforward in time. Thelen andhis colleagues have used thesesprinting simulations to com

pute the forces and work done by thehamstrings, and theyve also examinedhow small changes in the system mayaffect injury risk. Their results showthat strengthening core musclesattached to the pelvis, such as the hipflexors or abdominal obliques, havesubstantial influence on how the hamstring muscles are stretched and loaded

This suggests that exercises to improvecoordination and strengthen core mus-cles might reduce injury risk.Determining how individual musclesinfluence injury risk is very difficult toanswer experimentally, Thelen saysBut these simulations give us a way toestimate the effects.

NEUROMUSCULARAPPLICATIONS

Neuroprostheses for patients withspinal cord injury represent one of the

most dramatic success stories of computational biomechanics. For the past 30years, clinicians have performed experiments using electrodes to stimulate par-alyzed muscles in coordinated patternsTheir work has led to the creation ofimplanted functional electrical stimulation (FES) systems that allow partiallyparalyzed patients to hold a fork andfeed themselves, brush their teeth, orrun a comb through their hair.

Over time, biomechanical modelshave come to play a greater role in FESsystems. Basically, as the functions

that we would like to restore havebecome more and more complex, ourability to implement this by intuitionand trial and error is decreasing, saysRobert Kirsch, PhD, associate professor of biomedical engineering at Case

Western Reserve University and a member of the Cleveland FES CenterThats why a number of years ago Ibegan migrating toward the use of computational models to serve as a test bedfor potential interventions. He com

Freglys customized simula-tion suggested he modify his

walk in two ways: rotate his pelvisforward, and flex his knees a bitboth of which served to bring his

knees toward the midline of hisbody. After nine months of train-ing, Fregly managed a nearly 40percent reduction in a criticalmeasure for knee loading. Since previousstudies have suggested that the maximumreduction a high tibial osteotomy surgerycan offer is 50 percent, Fregly is encour-aged by these results, which have recentlybeen accepted for publication. Thesemodifications were completely intuitive inretrospect, of course. Fregly says. But noone had ever come up with them before

with just an experimental, trial-and-errorapproach. Now, hes conducting a pilotstudy of 10 patients with knee osteoarthri-tis to see if the same approach works forothers as well as it worked for him. Earlyresults are promising: The first subjectlearned the new motion in one trainingsession and managed a 45 percent reduc-tion in knee loading.

Fregly and his teams simulation andoptimized suggestions can be computedfor an individual in about a half-hour.The pilot study might help the

researchers determine whether everypatient needs to have a customizedsimulation built, or whetherFreglys simple gait sugges-tions are generalizable tomost knee osteoarthritispatients. Fregly eventual-ly hopes to bring this kindof individualized rehabili-tation to patients with a

variety of other conditionsas well, including patello-femoral pain and footulcers from diabetes.

To extend the train-ing rationale further,prevention of sports injuriesmight be an even more effectivestrategy than rehabilitation. Yetits not always clear exactly howand when injuries tend tooccur. Darryl Thelen, PhD,associate professor of mechani-cal engineering at University of

Wisconsin, uses computationalbiomechanics to analyze poten-

Thelen and his colleagues performed simulations of

sprinters in order to estimate hamstring muscle force,

work and fiber stretch during the swing phase of

sprinting. Courtesy Darryl Thelen.

Determining how individual

muscles influence injury risk

is very difficult to answer

experimentally, Thelen says.

But these simulations give us away to estimate the effects.



pares the high-stakes process to designin the aerospace industry. If yourebuilding an airplane, you test it in a

wind tunnel before you actually fly it.In our case, wed like to test many dif-

ferent FES systems before we actuallytry them in human subjects.

Since theres a limit to the numberof electrodes that can be implanted in apatient, surgeons need to choose their

muscles for implantation carefully.Several combinations of muscles areoften feasible candidates for implanta-tion. Kirsch and his colleagues can cus-tomize simulations of each patientsneuromuculoskeletal system so that cli-nicians can explore which system ofelectrodes and muscles would bringabout the greatest clinical benefit.

Kirsch, whose specialty is arm func-tionality, first measures the patientsstrength in various muscles and looksat restrictions in range of motion in the

elbow and shoulder. Then he and hiscolleagues fit a musculoskeletal modelof the upper arm to the patients indi-

vidual data. Like many otherresearchers, Kirsch didnt build thismodel from scratch; he works with anelbow-and-shoulder model originallydeveloped by Frans Van der Helm,PhD, at Delft University of Technologyin the Netherlands, which Kirsch hasadapted for particular use in spinalcord injuries. Kirsch says that the goal

of this work is not to build the mostdetailed possible model, but rather tocreate customized models that best pre-dict surgical outcomesand thus allowclinicians to circumvent the costly trial-

and-error process.Although individualized models

like these are immensely helpful forplanning patient treatment, clinicianscan also glean useful knowledge from

biomechanical models of an entiregroup of patients. Stroke survivors, forinstance, often displaying a ratherslow, asymmetrical shuffle. In order tobest rehabilitate these patients, clini-

cians would like to know exactly whatneuromuscular combinations are caus-ing these changes in the first place.JillHigginson, PhD, assistant professor ofmechanical engineering at University

of Delaware, is building musculoskele-tal models of both stroke survivors andhealthy older adults.

Higginson and her colleagues want tounderstand the muscle excitation pat-ternsstemming from faulty neural com-municationthat lead to the characteris-tic stroke gait. Using dynamic optimiza-tion techniques, theyre building simula-tions that will help highlight exactly

which patterns of muscle coordinationare common to stroke patients.Ultimately, we want to see what hap-pens if you change the muscular excita-tion pattern, Higginson says. We canrun that through the simulation.

The hope is that clinicians couldthen develop rehabilitation strategiesthat retrain the stroke patients neuro-muscular system. To that end, ThomasBuchanan is working with colleagues atthe University of Delaware to developtwo new types of therapy programs.One, developed with help from StuartBinder-MacLeod, PhD, a professor of

This schematic elucidates the primary algorithms used for computational biomechanics simulations across multiple scales. Courtesy of

Scott Delp and Clay Anderson.

If youre building

an airplane, you test

it in a wind tunnel

before you actually

fly it, says Kirsch.

In our case, wedlike to test many

different FES systems

before we actually

try them in

human subjects.



physical therapy, would use electrodes tostimulate particular muscles, stepping infor the damaged brain until it regainsmore function. For the other, SunilAgrawal, PhD, a professor of mechani-

cal engineering, is helping to designrobots to physically guide and reinforcepatients movements during physicaltherapy. Both approaches use computa-tional biomechanics methods to deter-mine which particular muscles mostneed support.

Biomechanical simulations are alsohelping guide individualized treatmentfor other disorders. People with cerebralpalsy, for instance, often display charac-teristic abnormal movements thatinclude stiff-knee gait, crouch gait, and

excessive hip rotations. Clinicians preferto intervene at a relatively young age,using a combination of strengtheningexercises, braces, casts, and in severecases, surgery. But common proceduressuch as detaching and rearranging ten-dons, or dividing and realigning bonesin the leg are extremely invasive and notalways successful.

Scott Delp and his colleagues aredeveloping musculoskeletal models ofcerebral palsy patients based on MRIscans and kinematic measurements.

The researchers want to identify the par-ticular biomechanical causes of walkingabnormalities in specific individuals.Once we know the underlying cause,

we can determine the best treatment,Delp says. They can then work with sur-geons to simulate the effects of differentkinds of surgery on an individualpatient, perhaps even devising new tech-niques. Recently, these cerebral palsymodels helped Delp and his colleaguesinvestigate ways to avoid a common

weakening of the calf muscle after cer-tain tendon-lengthening surgeries.

Simulations had suggested a specificmodification during surgery that mightimprove results. After turning to the lit-erature, the researchers discovered that50 years ago a similar surgical techniquehad been published but had been large-ly ignored. Delp and his team publishedtheir findings, including simulationresults and theoretical evidence.Surgeons are now reporting good resultsin treating cerebral palsy patients withthe rediscovered procedure.

With general

models, you can

uncover fundamental

principles, Delp says.

But its not clear to

which patients those

general principles

actually apply. We

need to customize

treatment for the

individuals.

A graphical representation of a shoulder and elbow musculoskeletal model, with mus

cles indicated by the red lines, and muscle wrapping surfaces represented by the blue

ellipsoid (thorax), sphere (humeral head) and cylinders (humerus, ulna and radius).

Courtesy of Dimitra Blana and Robert Kirsch.

CHALLENGESFOR THEFUTURE

At present, most clinical applicationsof computational biomechanics rely ongeneric models rather than fully cus-tomized simulations. Its a good firststep, but future treatment decisions willdemand greater specificity. With general models, you can uncover fundamentalprinciples, Delp says. But its not clearto which patients those general princi-ples actually apply. We need to cus-tomize treatment for the individuals.

With data from individualized simulations and treatment outcomesresearchers could then turn from purelydescriptive studies to analyses morefocused on predicting clinical outcomes

Researchers currently try to scale mus-

culoskeletal models to individual patientgeometry as much as possible, but fullindividualization on a large scale is stillprohibitive. It would be nice to have asimple way to adjust the models to theindividuals, Kirsch says. We can do itnow through painstaking manual meas-urements, but thats not practical for wide-spread deployment into the clinic. Someresearchers are working on ways to useMRI, CT and ultrasound data to morequickly and routinely scale flexible mus


10/10Wi t 2006/07 BIOMEDICAL COMPUTATION REVIEW 19bi di l p t ti i g

person. With new work in imaging, wecan see how a muscle is loaded duringfunctional movement, Thelen says. Alot of macro models dont capture thefull soft tissue complexity. Imaging cou-

pled with models could potentiallyimprove our representation and under-standing of soft tissue function.

As increasingly sophisticated bio-

mechanical models produce prodi-gious amounts of data, researchers willalso need to develop statistical meth-ods to help interpret the results. Likea lot of biology, computational biome-

chanics is overwhelmed by an incredi-ble wealth of data, Delp says. A sim-ulation can spit out 100 columns ofdata for each millisecond. We need a

way to quickly and objectivelyextract which factors are most

important therapeutically.As with any simulation, com-

putational biomechanical mod-els need to prove both their accu-racy and their worth. Modelersneed to work closely with experi-mentalists to demonstrate thattheir simulations are not onlyinternally consistent but alsoaccurately represent reality. Andas clinical studies increase innumber, researchers will gatherevidence showing whether the

use of simulations does in factimprove treatment outcomes.Improving clinical outcomes isour grand challenge, Delp says.Achieving this goal requiresthat experts in biomechanicalsimulation stay connected to therealities of the clinic.

Finally, researchers must tryto package simulations for every-day use and start to put them inthe hands of biologists and clini-cians. Collaborations are essen-tial, says Harry Hoyen, MD,

professor of orthopedics at CaseWestern Reserve UniversityMedical School, who works withRobert Kirsch on FES systems torestore function to the shoulderin partially paralyzed patients.The clinicians work and the sci-entists work go hand in hand.Our results are still relatively inthe beginning stages. But itsastonishing at present what wevebeen able to accomplish.

culoskeletal models to individual patients.These imaging techniques might also

prove useful in further refining existingmusculoskeletal models with soft tissuedynamics in vivo. Researchers have thus

far relied on static imaging work as wellas extensive analysis of cadavers. Butthese methods are no substitute forexamining soft tissue details in a living

The clinicians work and the scientists work go hand in hand.

Our results are still relatively in the beginning stages, says Hoyen.

But its astonishing at present what weve been able to accomplish.

Delp and Allison Arnold, PhD, have modeled the various abnormal gaits observed in patients with

cerebral palsy to help them evaluate appropriate treatments. Here, they used MRI images to create a

musculoskeletal model of an individual with a crouched internal-rotation gait. The rotational moment

arms of the medial hamstrings, adductors and other muscles were evaluated at the body positions cor-

responding to the subjects internally rotated gait. Photo courtesy of Eugene Bleck; images previous-

ly published in Arnold and Delp (2004), The role of musculoskeletal models in patient assessment and

treatment. In: Treatment of Gait Problems in Cerebral Palsy, Ed. Gage, MacKeith Press, London.