Design Case Studies in Rehabilitation Engineering
William DurfeeDepartment of Mechanical Engineering
University of MinnesotaMinneapolis, USA
And a few other projects
BRIDGE
• Stimulated Muscles = Power • Brace = Trajectory guidance • Brake = Control, stability
HUMAN/MACHINE DESIGN LABDepartment of Mechanical Engineering
University of Minnesota(www.me.umn.edu/labs/hmd/)
Fu
x,vT
X
�
�
PE Force-Velocity
CE Force-Velocity
Fscale
IRC
CE Force-Length
Activation Dynamics (2nd order)
PE Force-Length
u
V
X
V
X
Force
Passive Element
Active Element
Muscle mechanics
Smart orthotics + electrical stimulation for gait restorationHaptic interfaces for virtual product
prototyping, smart knobs for cars
Rehabilitation engineering-Tele-rehabilitation-Stroke rehab-Driving simulators
Human assist machines-Compact power sources-Powered exoskeletons-Natural control
Medical device design-Evaluation of surgical tools
www.dmdconf.org
Rehabilitation Engineering
The application of engineering principles to treatment, services and devices related to people with disabilities
20-49 million in U.S.
6.5% of GDP
Therapeutic Technology
Any equipment that treats an impairment
www.empi.com
www.donjoy.com
Assistive Technology
Any equipment that increases ability or function
bionics.ossur.de
www.abledata.com
www.abledata.com
www.dekaresearch.com
www.ric.org
www.technologyreview.com
Photos from Paraplegia News and Sports ‘n Spokes
MEDICAL VIEW
StrokeSCIMS
Muscular Dystrophy
ArthritisAmputationAlzheimer’s
Medical Condition Cognitive•Sensory/Motor integration•Memory•Reasoning
Physical
Motor
Sensory
•Speech•Balance•Gait•Coordination•Grip•Arm
•Sight•Sound
Impairment
PERSON VIEW
Person
Motor
Sensory
Cmd & Control
Hearing Sight
MobilityGrip
Arm•Wheelchair•Surgery•FES•Prosthesis•Orthosis
•Hearing aid•Cochlear implant•Signing
•Reading machine•Refreshable Braille•Glasses•Mobility
•ECU•Computer•Communic.•Robot•Speech•Driving
USER TASK
AT DEVICE
1
2
IT'S MORE THAN THE TECHNOLOGY
30-50% of AT devices are abandoned
TELEREHABILITATION
"Telerehabilitation is the clinical application of consultative, preventative, diagnostic, and therapeutic services via two-way interactive telecommunication technology."
American Association of Occupational Therapists Position Paper on Telerehabilitation
7 hrs
Why tele?
• Clients in rural locations• Clients in urban locations, but have
transportation challenges–No car–Poor public transportation
• Eliminates transportation time
TRAINING RECOVERY OF
HAND FUNCTION FOLLOWING STROKE
Collaborators: James Carey, Samantha Weinstein, Ela Bhatt, Ashima NagpalFunding: NIDRR, H133G020145
MOVEMENT
CONCENTRATION
LEARNING
Tracking task
HOME-BASED TRACKING
USABILITY
FeedbackTracking
CLINICAL TRIAL24 Subjects
2 to 305 miles from the U
One at 1,057 miles
180 trials/day x 10 days = 1800 trials
Pre-post function and fMRI
Task Variants
5, 10, 15, 20 secDuration
0-50%, 30-70%, 50-100%, 0-125% of
active range Amplitude
0.2, 0.4, 0.8 HzFrequency
Hand Position: Pronated, Mid, SupinatedJoint: Finger, WristHand: Ipsi, ContraVisual feedback: On, Off
Wave parameters
Wave shapes
100 combinations selected
Pre-Post Evaluations
• Box and Block• Jebsen Taylor Hand Function• Finger Range of Motion• Finger Tracking Performance• fMRI (cortical activation intensity and
location)
A. Box and Block, B. Jebsen Taylor, C. Finger ROM, D. AI score
Neurorahab Neural Rep, 21:216, 2007
Lesionon right
PRE
POST
Neu
rora
hab
Neu
ral Rep
, 21:2
16,
2007
Key Results
• Improved in tracking accuracy and finger ROM
• Improved on functional tests• Cortical activity shift towards lesioned side• Subjects had high tolerance for
technology, could self-install system and don/doff sensors
• Tracking and move group had similar results
Conclusion: Tracking training at home is feasible and effective.
NEXT STEPS
Practical Applications of Muscle Stimulation
• Bladder stimulation (incontinence)• Deep brain stimulation (movement
disorders)• Visual prostheses (artificial retina, cortical
stim)• Auditory prostheses (cochlear implant)• Pain suppression (TENS)• Pacemakers• Limb control (paralysis)
E-STIM APPLICATIONS
www.biotronik.de
Earl Bakken, University of Minnesota, Medtronic and the
portable pacemaker
medtronic.com and Bakken Museum
The Stimulator
Brain
Spinal Cord
Limb
Stimulator
Liberson foot-drop system, 1961
Heel switch triggered peroneal n. stimulationCorrection of foot-drop following strokeStarted field of FESSeveral commercial and research embodiments
Medtronic implanted foot-drop system
Upper Limb FES
Grasp restorationForearm and hand muscle stimulationRudimentary grip facilitates independence
NeuroControl FreeHand
Cleveland FES Center
Cleveland FES Center
FES-AIDED GAIT
FEXTERNAL
CONTROL STIMULATORInputs
Measurements
FEXTERNAL
• Improve health through weight bearing• Brief standing: social and functional• Limited ambulation in vicinity of wheelchair• No balance, no change in neuro function
Cleveland FES CenterCLEVELAND FES CENTER
PROBLEMS WITH FES-AIDED GAIT
•Not enough muscles•Not enough sensors•Low muscle forces•Muscle fatigues•No control over upper body•Muscles not electric motors•Size/weight/cosmesis
Requires precise, stable control for repeatable steps
Muscles are nonlinear,
time-varying
Need to walk reasonable distances
Muscles fatigue rapidly
STIMULATION PLUS SMART ORTHOTICS
Muscle stimulation
provides power
Brakes for locking and
control
Orthosis provides guidance
and support
INCREASED SPEED, DISTANCE
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
Without CBO With CBO
Gait Speed
0.09
0.12
Spee
d (m
/s)
0
10
20
30
40
50
60
Without CBO With CBO
Gait Distance
25
50
Dis
tanc
e (m
)
IEEE Trans Rehab Eng, 4(1):13, 1996, IEEE Trans Rehab Neural Eng, 2003
BETTER REPEATABILITY
0
20
40
60
80
100
120
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 20
20
40
60
80
100
120
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (sec)
With CBO
Time (sec)
Without CBO
Kne
e an
gle
(deg
)
Kne
e an
gle
(deg
)
IEEE Trans Rehab Eng, 4(1):13, 1996, IEEE Trans Rehab Neural Eng, 2003
NEXT GENERATION
Prostheses/Orthoses Complexities
• Intimately coupled with body• High forces• Power transmission
Prosthesis (Gk: “prostitheni”, to add)Orthosis (Gk: “orthos”, straight)
• Fitting– Spread the load– Zero shear– Maintain blood flow
I.D. Magazine May 1998
Hip belt
Placeholders for brakes
Knee brace
Prismatic joint
Ab/adductionhinge
Medial hinge
ENERGY STORING BRACE
Energy Budget• ~30 nM over 60 deg
of motion• 31.4 J per extension• Extract 14 J per cycle
J. Biomechanical Engineering, 127(6):1014-1019, 2005.
ADAMS dynamic model
Gas springs
Cylinders
Accumulator
Non-invasive diagnostics of muscle activity
Mechanical and electrical properties provide window into muscle excitation contractionSmart stimulation, smart system id can isolate subsystemsFor differential diagnosesGo beyond, “This muscle is weak, let’s biopsy.”
MUSCULOSKELETAL SYSTEMSTIM
u(t)
FORCE/MOTION
y(t)
Model OverviewStimulation in, joint angle out. Muscle acts on skeletal system to product measurable joint motion. Model must account for muscle dynamics, joint geometry and limb dynamics.
MUSCULOSKELETALSYSTEMSTIM
u(t)
FORCE/MOTION
y(t)
MT GEOM
GEOM-1
u(t)
LDθ(t)τ(t)f(t)
x(t)
MUSCLES
Force = f(neural input, length, velocity, time, ...)
F
Activation
F
Velocity Time
FF
Length
a(t)LM(t)
VM(t)
CE
SE
PE
muscle
tendon
FM(t)=FT(t)=f(t)
passive
WHAT'S WRONG WITH THE MUSCLE MODEL
Invariant F-A, F-L, F-V (no change with activation)Invariant twitch dynamics (uniform fiber types)Time-invariant (no fatigue)
CEKSE
XCEXSE
XMT
Skeletal muscle, isometric twitch
MODELING ISOMETRIC MUSCLE
Staticnonlinearity
Lineardynamic system
Hammerstein model
stim force
Identify LDS with impulse responseDurfee & Palmer, IEEE TBME, 1994
Identify SL by deconvolutionDurfee & MacLean, IEEE TBME, 1989
020406080
100120140
0 100 200 300 400
Time (mS)
Torq
ue
2)( ask+
SINGLET
BEST FIT
Linear system fit to singlet
Muscle Force-Velocity
FORCE
VELOCITYSHORTENINGLENGTHENING
LINEAR FIT
MODEL OK FOR ISOLATED MUSCLE
0
5
10
15
20
25
30
35
0 4 8 12 16
Forc
e (N
)
Time (s)
IEEE Trans. Biomed. Eng., 41(3):205, 1994)
Experiment
Model
Using a nonlinear muscle behavior for
a diagnostic
0
50
100
150
200
250
300
350
400
450
500
0 50 100 150 200 250 300 350 400
Time (ms)
Torq
ue
Single pulse twitch
0
50
100
150
200
250
300
350
400
450
500
0 50 100 150 200 250 300 350 400
Time (ms)
Torq
ue
Double pulse twitch, if ideal linear system
y 2y
0
50
100
150
200
250
300
350
400
450
500
0 50 100 150 200 250 300 350 400
Time (ms)
Torq
ue
Double pulse twitch, real
3.2 y
Experiments
F-L, F-V, contraction dynamics, doublet propertiesBuild database for several muscles in non-impaired subjects
Nash Avery Search for Hope Fund via the Paul and Sheila Wellstone Muscular Dystrophy Center, University of Minnesota.
Force vs. Acceleration
-15-10-505
1015
-2 -1 0 1 2
Acceleration (g)
Forc
e (lb
-f)
Estimating limb inertia, simple clinical tool
POWEREDHUMAN-ASSIST
TOOLS
Engineering Research Center for Compact & Efficient Fluid Power
•Compact power sources•Natural interface and control
•Portable and/or wearable
BENCHMARKING ACTUATORS
And a few other projects
MUSCLE!
MUSCLE
• 40% of body weight• 640 move you• Work in pairs, can only push• Eye muscles move 100,000
times/day• Gluteus maximus is largest• Sartorius is longest
Muscle metrics
• Short-stroke, linear actuator– 5-20% shortening stroke
• Pull force: 30 lbs/sq. in.• 90 W/lb
– 180 lb athlete w/ 72 lb of muscle puts out 370 W sustained 5 W/lb for human muscle for continuous use
• 25% efficient• Compliant, back-drivable• Fatigues• Clean• Quiet !
Vogel (2001), "Prime Mover"Vogel (2001), "Prime Mover"
Power (W/lb)
0
50
100
150
200
250
Muscle--peak Muscle--sustained
Electric motor Automobileengine
Vogel (2001), "Prime Mover"(Aircraft engine, piston: 700; Aircraft engine, turbine: 2500)
Miniature free-piston air-compressor
FUTURE MICRO FPAC + 1KPSI TANK