Pravin Nair December 12, 2003 Slide 1 of 30 DEVELOPMENT OF QUANTITATIVE MEASURES FOR CHARACTERIZATION OF UPPER LIMB DYSFUNCTION Pravin Nair Advisor : Dr.

Pravin NairDecember 12, 2003Slide 1 of 30

DEVELOPMENT OF QUANTITATIVE MEASURES FOR CHARACTERIZATION OF UPPER LIMB DYSFUNCTION

Pravin Nair

Advisor : Dr. Venkat Krovi

Mechanical and Aerospace Engineering Department

State University of New York at Buffalo.


Presentation Overview

Motivation- Stroke, Rehabilitation and Diagnosis- Research Goals

Background- Robotic Therapy Devices

Implementation Framework- Hardware and Software Integration- Parameterized Exercise Protocols

Experiments - Design of Experiments

Results and Analysis- Mathematical Preliminaries- Quantitative Measures- Analysis of Obtained Data

Conclusions and Future Work - Summary- Work in Progress


Stroke and Rehabilitation• when a blood clot blocks a blood vessel or artery, or when a

blood vessel breaks, interrupting blood flow to an area of the brain, causing brain cells to die.

• Each year, over 750,000 people experience a new or recurrent stroke, leading to motor disability and upper limb (UL) dysfunction [NSA].

• Rehabilitation A goal-oriented process, which enables individuals with impairments to reach their optimal physical, mental and/or social functional level.

• Functional recovery linked to the duration, frequency, regularity and intensity of the rehabilitation therapy [1-5].

• Diagnosis A term which names the primary dysfunction towards which the therapist directs the Rehabilitation regimen

Motivation Background Implementation Framework Experiments Results & Analysis Future Work

STROKE

Rehabilitation

References [1-5] listed in slide 32

Diagnosis


Rehabilitation Regimen Implementation Issues

1. Careful characterization of the functional impairment.

• Variability within population

• Variability due to disease progress

• Subjective v/s Objective Assessment

2. Overall economic viability and logistics of deployment.

Accurate and Ongoing Assessment

Infrastructure Access/Costs v/sFunctional Recovery

Logistics:• Inpatient• Outpatient• Home-based

Exercise Regimen:• Free-Motion• Machine-Assisted



Goal of the Research Work

• A low-cost, home-based diagnostic and rehabilitation tool.

• Implementation as an immersive Personal Movement Trainer:

• Adequacy for quantitative assessment and ability to differentiate between users.


Target Audience: People with Upper Limb (UL) dysfunction due to Stroke

COTS + Virtual Environment + Path Devices Library

Virtual Driving Environment


Constraint-Induced Therapy

Constraint Induced Therapy The patient’s less impaired arm is restrained, and the patient intensively practices moving the more impaired arm, with feedback from a therapist.

• Improves functional use

• Expands cortical representation of the exercised limb.

• Needs continuous monitoring

• Needs specialized equipment or crude methods for restraining the less impaired limb

Advantages:

Disadvantages:

Constraint Induced Therapy



Current Technology and their Shortcomings

1.Current functional assessment (diagnostic testing) is subjective or semi-quantitative

2. Existing Robotic Therapy Devices (low-cost, portable, force-feedback devices) are specialized and concentrate on rehabilitation (as opposed to diagnosis and rehabilitation).

• MIT-MANUS• Rutgers Master II (RMII)• ARM Guide, JAVA Therapy• PHANTOM-based diagnosis

Examples of existing Robotic Diagnosis and Rehabilitation devices



Existing Specialized Robotic Therapy Devices

MIT-MANUS [1] Rutgers Master II (RMII) [2]

ARM Guide [3]

[1] M. Aisen, H. Krebs, N. Hogan, F. McDowell, and B. Volpe. “The effect of robot-assisted therapy and rehabilitative training on motor recovery following stroke,” Arch . Neurol., vol. 54, pp. 443–446, Apr.1997.

[2] V. Popescu, G. Burdea, M. Bouzit, and V. Hentz. “A Virtual-Reality-Based Telerehabilitation System with Force Feedback,” IEEE trans. on Information Technology in Biomedicine, vol. 4, no.1, March 2000. [3] D. Reinkensmeyer, B. Schmit, and W. Rymer, “Assessment of active and passive restraint during guided reaching after chronic brain injury,” Ann. Biomed. Eng., vol. 27, pp. 805–814, 1999.



“JAVA Therapy” COTS device [1]

Existing COTS-device and a Specialized Diagnostic Tool

PHANTOM-based diagnostic tool [2]

[1] D. Reinkensmeyer, C. Painter, S. Yang, E. Abbey, and B. Kaino, “An Internet-Based, Force-Feedback Rehabilitation System for Arm Movement after Brain Injury,” Proceedings of Technology and Persons with Disabilities Conference, 2000.

[2] A. Bardorfer, M. Munih, A. Zupan, and A. Primožič. “Upper Limb Motion Analysis using Haptic Interface,” IEEE/ASME transactions on Mechatronics, Vol.6, No.3, September 2001.



Our Hypothesis

Emphasizes:• COTS force-feedback devices• Immersive environment• Parameterized exercises

At Present:

1. One such device developed

2. Diagnostic capabilities only

GAMING DEVICES

Functional Assessment &

Motor Rehabilitation

InterfaceHardware

EXERCISE-PROTOCOLS

Low-Cost Mass Produced

Devices

Immersive Virtual

Environment

ParameterizedTherapies

IMPLEMENTATION FRAMEWORK

DESKTOP/LAPTOP PC



Overall Implementation

User Input

pedals

Vehicle Visualization

Immersive Driving Scenario

User Input from wheel &

pedals

Vehicle Kinematics

Parameterized Exercise Routines

Paths parameterized by amplitude and frequency



Virtual Vehicle Model

Knife edge kinematic model of a differentially driven wheeled vehicle.

Motion of the origin of the body-fixed reference frame w.r.t. the inertial frame:• Vx in the body-fixed x-direction• Vy in the body-fixed y-direction ( Vy = 0)

• angular velocity ω

The user is considered to be driving a differential-drive vehicle which can be modeled using the knife-edge model.

.

.

.

.

.

cos 0

sin 0

0 1

1/ / 2

1/ / 2l

r

x

yV

R Rb

R Rb



Simulink Implementation of the System

Simulink Block diagram of the system implemented

Data Collection

Vehicle Kinematics

Data Output



Visualization Aspect of the system

Example of a 2D GUI which allows conduct of the experiment and provides immediate relevant statistical feedback.

Examples of 3D visual interfaces for our Virtual Driving Environment (a) with simple parametrically generated paths; (b) with realistic roads from a database.



Testing Procedure

1. Subjects are “healthy”

2. Arm angles fixed at:

θ1 = 45°

θ2 = 60°

3. D1 and D2 adjusted to maintain a fixed offset

Assumptions and Standards :

Experimental Test Setup,

Schematic with relevant parameters



Guide the “vehicle” along parametrically generated paths, remaining as close as possible to the center line, with 3 preset forward speeds.

TASK

Testing Procedure (Cont:)

2D GUI Patient/Therapist Interface

Parametric library of labyrinthine maze-style paths and sinusoidal paths used for the diagnostic testing routine



x-coord. y-coord.

0.0025109 0.0747880.003825 0.0821720.005139 0.0895560.0065688 0.0969180.0079985 0.104280.0094857 0.111630.01103 0.118970.012632 0.12630.014263 0.13362

-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.60

0.5

1

1.5

2

2.5

3

3.5

4

Desired Path

Actual Path

Sample collected data

Plot of the user generated and actual paths

Movie file depicting the testing routine

Testing Procedure (Cont:)



Planar Curve and Cubic Spline

2 3( )i i i i iy t a b t c t d t

ˆ ˆ( )t x t i y t j R

0

( ) ( ) ( )t

t

s t t t dt R R

( ) ˆd s

ds

RT

ˆˆd

ds

TN

( )( ) d dTR s

ds dss

Planar Curve

Arc Length

Curve Tangent

Curvature of a

Curve

Curve Normal

Cubic Spline

X

Y

R(t)

T̂

N̂

Where, (0)

(0)

3( (1) (0)) 2 (1)

2( (0) (1)) (1)

[0,1], 0,1,...,

i i

i i i

i i i i i

i i i i i

a y

b y D

c y y D y

d y y D y

t i n

Where, is the tangential angle



0 5 10 15 20 25 30 35 40 45 50-10

-5

0

5

10Signal Corrupted with Zero-Mean Random Noise

time (seconds)

0 50 100 150 200 250 300 350 400 450 5000

50

100

150Frequency content of y

frequency (Hz)

Fourier Mathematics

)1( ,)(1

][)1)(1(2

1

NjekZN

jz N

kjiN

k

)1( ,][]][[][)1)(1(2

1

NkejzjxDFTkZ N

jkiN

j

N

k

N

j

kZN

jz1

2

1

2 ][1

][

Original Periodic Signal

Discrete Fourier Transform

Spectral Energy



-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.60

0.5

1

1.5

2

2.5

3

3.5

4

Desired Path

Actual Path

Performance MeasuresA quantity which explicitly expresses some desirable characteristic of an individual which helps in categorization of the (motor) ability/skill of that individual.

1. Error Value Parameter (EVP)

Difference between the desired and actual path at each time instant.

Error Value Parameter (EVP)Curvature-Based Performance Measure

Discrete Fourier Transform-Based Error Measure

Our Measures



Principal Harmonic

0 5 10 15 20 25 30 350

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

Frequency

Ma

ga

nit

ud

e

Frequency Spectrum

Secondary Harmonics

Performance Measures (cont:)

2. Curvature-Based Performance Measure

Comparison between the desired and actual path curvatures at for the corresponding arc length.

3

2 2 2

' '' ' ''

' '

x y y x

x y

2. Discrete Fourier Transform-Based Error Measure

Measure

UsefulEnergyDFT

TotalEnergy

2

2

1

[ ]

1[ ]

k pMeasure N

k

Z k

DFTZ k

N

2 3

2 3

x x x x

y y y y

x a b t c t d t

y a b t c t d t

Where,

Or,



1 2 3 4 50

2

4

6

8

10

12

Subject

Tim

e (

Se

c)

Time taken to complete the task

LowMediumHigh

Result Analysis

0 2 4 6 8 10 120

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Time (sec)

Err

or

Va

lue

Pa

ram

ete

r (E

VP

)

Time v/s Error Value Parameter (EVP)

data1data2data3data4data5data6

Subject 4

Subject 2

Subject 1

Subject 3

Subject 5

Plot showing the EVP plotted against the time value collected from the subjects for ‘Sine1’ at ‘Low’ speed.

Graph showing time taken by the subjects to traverse the path ‘Sine1’ at all speeds.



0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5-4

-3

-2

-1

0

1

2

3

Arc Length

Cur

vatu

re

Plot of Curvature v/s Arc Length for all Subjects superimposed over the Expected Curvature

Subject 4

Subject 2

Subject 1

Subject 5

Expected Curvature

Subject 3

Results (cont:)

Plot of curvatures of the user generated curve superimposed over the expected curvature.



Results (cont:)

0 1 2 3 4 5 6 7 8 9 100

0.05

0.1

0.15

0.2

0.25

0.3

Frequency

Mag

nitu

de

Frequency spectrum of all users traversing 'Sine1' path at a 'Low' speed

Subject 4

Subject 2

Subject 1

Subject 5

Subject 3

Subject No. Energy Ratio

1 0.2364

2 0.2297

3 0.2424

4 0.2355

5 0.2575

Plot of Frequency spectrum of user generated curves at ‘Low’ speed on ‘Sine1’ path.

Table of Energy Ratios



Results (cont:)

0 2 4 6 8 10 120

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Time (sec)

Err

or V

alue

Par

amet

er (

EV

P)

Time v/s Error Value Parameter (Subject 2)

Trial 2

Trial 1

0 2 4 6 8 10 120

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

Time (sec)

Err

or V

alue

Par

amet

er (

EV

P)


Trial 1

Trial 2

0 2 4 6 8 10 120

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

Time (Sec)

Err

or V

alue

Par

amet

er (

EV

P)

Time v/s Error Value Prameter (Subject 3)

Trial 1

Trial 2

0 2 4 6 8 10 120

0.2

0.4

0.6

0.8

1

1.2

1.4

Time (sec)

Err

or V

alue

Par

amet

er (

EV

P)


Trial 2

Trial 1

Plots of comparison of subject performance (Error Value Parameter) in two trials- Trial 1: Initial test and Trial 2: Test after 6 months.



Results (cont:)

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5-2.5

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

2.5Curvature v/s Arc Length (subject 1)

Arc Length

Cur

vatu

re

Trial 1

Trial 2

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5-4

-3

-2

-1

0

1

2

3Curvature v/s Arc Length (subject 2)

Arc Length

Cur

vatu

re

Trial 2

Trial 1

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5-1.5

-1

-0.5

0

0.5

1

1.5

2Curvature v/s Arc Length (subject 3)

Arc Length

Cur

vatu

re

Trial 2

Trial 1

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5-2.5

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

Arc Length

Cur

vatu

re

Curvature v/s Arc Length (Subject 4)

Trial 2

Trial 1

Plots of comparison of subject performance (Curvature Comparison) in two trials- Trial 1: Initial test and Trial 2: Test after 6 months.



Conclusions

Considerable promise for improving the speed, resolution and quality of diagnosis.

Successful development, implementation and testing of a low-cost diagnostic tool

Preliminary tests display the potential of the set-up as a diagnostic tool



Continuing/Future Work

Test the tool with people with UL dysfunctions

Provide force-feedback to the subjects

Develop a 3-D interface as patient feedback to the therapist

Add strength training/recovery aspects to the tool

Network the tool through the internet



Questions ?


Thank You


References

1. M. Aisen, H. Krebs, N. Hogan, F. McDowell, and B. Volpe. “The effect of robot-assisted therapy and rehabilitative training on motor recovery following stroke,” Arch. Neurol., vol. 54, pp. 443–446, Apr.1997.

2. H. Krebs, N. Hogan, M. Aisen, and B. Volpe. “Robot-Aided Neurorehabilitation,” IEEE transactions on Rehabilitation Engineering, vol. 6, no. 1, March 1998.

3. B. Volpe, H. Krebs, N. Hogan, L. Edelsteinn, C. Diels, and M. Aisen. “Robot training enhanced motor outcome in patients with stroke maintained over 3 years,” Neurology, vol. 53, pp. 1874–1876, 1999.

4. K. Kwakkel et al. “Effects of intensity of rehabilitation after stroke, a research synthesis,” Stroke, vol. 28, no. 8, pp. 1550–1556, 1997.

5. P. Langhorne, R. Wagenaar, and C. Partridge. “Physiotherapy after stroke: More is better?,” Physiotherapy Res. Int., vol. 1, pp. 75–88, 1996.


Curvature Comparison Example Example

2

2 2 2

ˆsin( )( )

1 cos ( )

aw t iR s

a w t

sin( )x a t y t

ˆ ˆ( ) sin( )R t a t i t j

ˆ ˆ( ) cos( )R t aw t i j

( )s t R R dt

2 2 2( ) ( ) cos ( ) 1R t R t a w t

0

2 2 2( ) 1 cos ( )t

t

s t a w t dt

2 2 2( ) 1 cos ( )ds

s t a w tdt

2 2 2

ˆ ˆ( ) cos( )( )

( ) 1 cos ( )

dR dR dt R t aw t i jR s

ds dt ds s t a w t

( )R s

2

2 2 2

sin( )

1 cos ( )

aw t

a w t

Motivation Background Performance Measures Implementation Framework Experiments & Results Future Work


Discrete/Fast Fourier Transform Analysis

Principal Harmonic

0 5 10 15 20 25 30 350

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

Frequency

Ma

ga

nit

ud

e

Frequency Spectrum

Secondary Harmonics

Frequency Spectrum of a signal

_UsefulEnergy

FFT MeasureTotalEnergy

Where, Useful Energy is the energy captured by the more significant peaks of the Fast Fourier Graph.



Continuing Work



Detailed Simulink Block Diagram


Functional Interaction

Therapist’s Interface

Virtual Environment

Graphical Display

Therapist

Device Manager

Subject

Internet

User Interface Internet

Functional interaction of the visualization, programming and data acquisition components.


Mean Deviation Result Table

Mean deviation values for the subjects at all the 3 speeds for the 3 sinusoidal paths.


Pravin Nair December 12, 2003 Slide 1 of 30 DEVELOPMENT OF QUANTITATIVE MEASURES FOR CHARACTERIZATION OF UPPER LIMB DYSFUNCTION Pravin Nair Advisor : Dr.

Documents

rehabilitation therapy

analysis future work

progress slide

rehabilitation tool

future work summary

quantitative assessment

stroke cots

forcefeedback devices