458 Copyright © 2008 The American Society of Neurorehabilitation

Motor Imagery to Enhance RecoveryAfter Subcortical Stroke: Who Might Benefit,

Daily Dose, and Potential Effects

Lucy Simmons, BSc, Nikhil Sharma, MB, ChB, MRCP(UK), Jean-Claude Baron, MD, FRCP,FMedSci, and Valerie M. Pomeroy, PhD

Background. Motor imagery may enhance motor recovery afterstroke. Objectives. To estimate the proportion of patients able toperform motor imagery, the feasibility of delivery of motorimagery training (MIT), and the effects of MIT on motor recov-ery in an exploratory study. Methods.An immediate pretreatmentand posttreatment single-group design was used to study 10patients after subcortical stroke with neuromuscular weakness inthe upper limb. MIT that included upper limb activities reflect-ing everyday tasks was provided for 10 consecutive working days.Measures included assessment of chaotic motor imagery, patientreport of tolerability of MIT, Motricity Index (MI), Nine HolePeg Test (9HPT), and quality of movement (MAL-QOM). MITdose was changed in response to patient feedback. Graphedmotor function scores were inspected visually for clinicallyimportant changes. Results. Four of the 10 patients were unable toperform motor imagery. Patient opinion was positive about thecontent and shaped daily dose of MIT given in two 20-minuteperiods separated by a 10-minute rest. Clinically importantchanges in motor scores were found. Four patients increased MIscore (range 8-16), 3 patients increased 9HPT score (range 0.02-0.04 pegs/second), and 4 patients increased MAL-QOM score(range 0.63-1.29). Conclusions. MIT was received positively bypatients, but 40% were unable to perform imagery and interindi-vidual variations were found on motor function.

Key Words: Stroke—Rehabilitation—Physical therapy Motorimagery.

An appropriate dose of functional activities mightenhance upper limb recovery after stroke.1,2 Ifpatients have sufficient voluntary activation of

muscles to produce movement, they can participate inrepetitive, goal-directed functional activities. If, how-ever, patients do not have sufficient voluntary activationof muscles, the challenge is how to provide such task-related input to facilitate activation of muscles in afunctionally related temporal-spatial sequence.

A therapy based on motor imagery may be a way for-ward as, in principle, imagery is not dependent on resid-ual function. Imagery does incorporate voluntary drive,3

which is important for motor learning after neurologicaldamage.4 Furthermore, motor imagery has been shownto access the cortical motor network after stroke5 andincrease excitability of appropriate spinal reflex path-ways.6,7 Motor imagery (reviews8,9) can be defined as “adynamic state during which the representation of a spe-cific motor action is internally reactivated within work-ing memory without any overt motor output, and that isgoverned by the principles of central motor control” and“occurring from the first person perspective.”10

Consequently, an observer cannot judge whether anindividual is engaged in motor imagery or is using analternative strategy such as visual imagery, that is, fromthe third-person perspective or indeed is engaged in anymental representation of an activity. Assessing compli-ance with motor imagery training or therapy (MIT) ismade even more complex as the stroke itself may dam-age the intrinsic ability to perform motor imagery orresult in temporal uncoupling (ie, no longer restrictedby the principles of motor control8). We have labeledthese 2 possibilities chaotic motor imagery.10 However,the characteristics of chaotic motor imagery and itsimplications on recovery need further investigation.

Motor imagery has been recognized as a physicaltherapy technique for a considerable period of time(eg,11) and is already included in the conventional ther-apy package provided for people after stroke by physio-therapists in the United Kingdom.12 Nevertheless, theevidence base for the use of MIT to enhance motorrecovery after stroke is not compelling. A systematic lit-erature search to identify papers that reported investiga-tion of motor imagery to enhance upper limb recoveryafter stroke identified 5 studies.10 Although all 5 studiesreported positive effects of motor imagery on motorfunction, these results need to be interpreted with refer-ence to the facts that none of these studies undertook acomprehensive assessment for chaotic motor imagery withpatients, and only 2 studies specifically instructed patientsto perform motor imagery. Interpretation of these studiesis also limited by small sample sizes, omission of patients

From the University of Cambridge (LS, NS, J-CB), and St George’sUniversity of London (now University of East Anglia) (VMP).

Address correspondence to Professor Valerie M. Pomeroy, PhD,Professor of Neurorehabilitation, University of East Anglia, QueensBuilding, Norwich, NR4 7TJ, UK. E-mail: v.pomeroy@uea.ac.uk.Neurorehabil Neural Repair 2008:22:458–467

DOI: 10.1177/1545968308315597

with severe paresis (arguably those who could benefitthe most from this priming technique), heterogeneity ofthe motor imagery interventions, and the outcomemeasures employed.

This pilot study of MIT aimed to estimate the propor-tion of patients with subcortical stroke who performchaotic motor imagery; explore the feasibility of MIT interms of patient tolerability of daily dose and content; andexamine the effects of MIT on upper limb strength, handdexterity, quality of movement, and functional ability.

METHODS

Design

A pretherapy, posttherapy, single-group design wasused for this feasibility study. Ethical approval was givenby the Cambridge Regional Ethics Committee. Eligibleparticipants were 20 to 85 years old; had a subcortical,first-ever stroke (infarct or hemorrhage) with neurolog-ical deficit in the affected upper limb; were right-handedas assessed by the Edinburgh Handedness scale13; had nohistory of significant cognitive impairment, aphasia, orunilateral inattention; had no active malignant disease orrenal, liver, or cardiac failure; had no other neurologicaldisease or depression; and were not receiving benzodi-azepines or antidepressant medication. In addition,patients had to exhibit the ability to perform motorimagery (see measurement battery section).

Patients were treated at the hospital clinic or homeby patient choice, after being recruited by a researchphysiotherapist from the inpatient stroke unit and theoutpatient stroke clinic of a teaching hospital.

Procedure

Patients who passed the first part of the ChaoticMotor Imagery Assessment (CMIA) and gave informedconsent were included. All patients were then assessedfor ability to perform motor imagery by completingparts 2 and 3 of the CMIA. Those who were unable toperform motor imagery (4 of 10 patients) wereexcluded from the intervention part of this study.Patients who were able to perform motor imagery thenundertook the remainder of the measurement battery(baseline measure point) before beginning the interven-tion on that day or the following day. Intervention wasgiven each working day for 2 weeks (10 days). Patientsundertook the measurement battery, except for theCMIA, on the day of the last treatment or the followingday (outcome measure point). Follow-up measureswere made using the measurement battery, except for

the CMIA, 4 to 8 weeks after completion of the lasttreatment session.

Intervention

For each MIT session, patients sat in a chair at a tablethat allowed a position as close as possible to:

• hips, knees, and ankles at 90 degrees, with head helddirectly over pelvis with spine erect

• forearms resting on the table without elevation/depres-sion of the shoulder girdle

• during the motor imagery training, patient adopted ahand position (forearm either pronated, mid position,or supinated) appropriate to the task to be imagined

On the first day of training, the rules of motor imagerywere explained as:

• each action is imagined in the first person, that is,patients imagine they are moving

• imagining that another person is doing the action(visual imagery) should not be practiced

• during motor imagery, the patient should not makemovements

The instruction given to patients before each sessionwas, “During this session there are some activities thatyou are going to imagine doing with your [researchphysiotherapist indicates the paretic upper limb]left/right arm.” For each activity:

1. The physiotherapist demonstrated the movement tobe imagined by the patient: 2 repetitions

2. Patients use the nonparetic upper limb to physicallyperform the movement: 2 repetitions

3. Patients imagine the movement using the nonpareticlimb: 2 repetitions. The instructions given were, “Closeyour eyes. Concentrate on your hand just resting onthe table but do not move it. Concentrate on how itfeels just resting there. Do not move your fingers, handor arm. Now imagine ………. [individual activitydescribed here]. Just imagine it and do not move any-thing. Open your eyes when you have done this actiontwo times.”

4. Patients imagine movement using paretic upper limb:3 repetitions (repeated 3 times to provide a total of 9repetitions of the imagined activity). Same verbalinstructions given as for nonparetic limb.

The physiotherapist checked for actual movement andintermittently asked patients to describe what they weredoing and what they were experiencing. An example ofthe type of question asked is,“Can you tell me what is hap-pening when you are doing this?” The physiotherapist alsoasked specifically whether the patients were imagining

Motor Imagery Poststroke: Dose and Aptness

Neurorehabilitation and Neural Repair 22(5); 2008 459

Simmons et al

460 Neurorehabilitation and Neural Repair 22(5); 2008

A. Isolated functional movements1. Flexion/extension of thumb (hand in pronation)2. Flexion/extension of index finger (hand in pronation)3. Circling of thumb4. Circling of index finger5. Abduction/adduction of all digits (spreading of hand – sideways)6. Making a fist/spreading hand 7. Moving extended fingers backwards and forwards8. Moving hand between ulnar and radial deviation9. Circling of the wrist

10. Wrist flexion/extension11. Moving between pro/supination of the hand12. Flexion/extension of the elbow in supination13. Flexion/extension of the elbow in mid position 14. Flexion/extension of the elbow in pronation15. Elevation/depression of the shoulder girdle (shrugging)16. Shoulder girdle circumduction17. Shoulder flexion/extension 0-90° of shoulder flexion18. Shoulder abduction/adduction 0-90° with elbow flexed19. Lateral/medial rotation of the shoulder (moving hand in an arc on the table)20. Lateral/medial rotation of the shoulder with humerus in 90° abduction (slowing traffic)

B. Combined functional movement – hand focused21. Trace shape of a triangle with index finger22. Trace shape of a square with index finger23. Opening up elastic band – round fingers and thumb24. Rolling plasticine into a sausage on the table25. Rolling plasticine into a ball on the table26. Flicking movement27. Beckoning with index finger28. Making the “OK” sign then hand back to table29. “Thumbs up” sign and then hand back to table30. “Fingers crossed” sign31. Scissors action with digits 2 and 332. Scratching paint off the table33. Winding wool round a short vertical pole34. Stirring water in a bucket with your index finger35. Pouring sand from cupped hand onto table36. Picking up buttons and dropping them into a saucer37. Turning a doorknob38. Knocking on a table39. Slapping your hand on a table with arm static40. Turning cards over

C. Combined functional movement – arm and hand focused41. Taking soup from hand to mouth in a spoon42. Pick up a cup, to mouth to drink, cup to table, let go43. Waving to someone nearby44. Waving to someone far off (over head)45. Stirring a tin of paint46. Pick up phone handset, place to your ear (ipsilateral), and then put it back on the table47. Brushing/combing hair48. Holding stick – trace a zig-zag in the air49. Pulling a drawer open50. Knocking on a door51. Painting a wall with a paint brush52. Stretching arm and hand up (waking up)53. Cleaning windows54. Cleaning table55. Crawling up wall with hand56. Touching top of head57. Hand behind head58. Hand behind back59. Touching nose, eyes, ears60. Straightening arm out to the side

Box 1. Tasks used for motor imagery training.

Motor Imagery Poststroke: Dose and Aptness

Neurorehabilitation and Neural Repair 22(5); 2008 461

themselves performing the action or whether they wereseeing the action occur as a passive spectator.

The tasks used for MIT were designed to provide arange of upper limb functional activities, which reflectaspects of functional actions used in everyday activity.The 3 types of activity used in motor imagery trainingare shown in Box 1.

These activities were performed in a random order,which was preset by a computer program. This methodwas chosen to provide an equal amount of practice ofeach individual action, to prevent an order of perfor-mance effect (eg, the first activities may be performedwith more vigor than the last), and to provide enoughvariation in each session to maintain motivation andcompliance.

The 10 sessions were planned to last for 70 minuteswith two 30-minute periods of MIT separated by a 10-minute rest. This was amended iteratively in response topatient feedback (see measurement battery and analysisfor detail).

Measurement Battery

1. Chaotic Motor Imagery. Ability to perform motorimagery was assessed using the CMIA, described indetail elsewhere5 and summarized in Appendix 1. TheCMIA consists of 3 tests performed in this order:intrinsic ability to perform motor imagery, activelyengage and maintain the motor imagery network, andnot use alternative strategies. First, patients are shown96 A4-sized picture cards of hands (4 different views,12 rotations, left and right) and asked to identifywhether the picture is of a left or right hand. A scorebelow 75% correct indicates that the patient is unableto perform accurate motor imagery. Second, patientsare asked to perform motor imagery of a paced fingersequence task (2,3,4,5,2) using auditory cues at 1 Hz.They first execute the movement and then try motorimagery between a block of rest. Patients were askednot to view the activity from the third person or toassign numbers or tones to each finger. To monitorcompliance during motor imagery, the duration of thefinger tapping exercise was varied and the patients hadto confirm the position at the end of each block bytouching the appropriate finger. Third, patients per-form the same finger-tapping sequence initially usingexecuted movement and then using motor imagery.Auditory pacing varies, however. The rate starts at 40beats/min and increases by 10 beats every 5 seconds.The break point is defined as the time when the patientis unable to perform the task accurately. Patients areexcluded if the break point is greater for motorimagery than for executed movement.

2. Patient tolerability of dose and content of MIT wasassessed by the physiotherapist obtaining feedbackfrom patients during and at the end of each MIT session.

During each treatment session, any comments madeby patients were noted down.

3. The Motricity Index (MI) is a clinical measure of abilityto produce a voluntary contraction of muscle (strength).The total possible score for the arm subsection is 100,with a higher score indicating better function.17

4. Effect of MIT on hand dexteritya. Nine Hole Peg Test (9HPT). Patients to pick up from a

container 9 wooden dowels (9 mm diameter by 32 mmlong) and then place them into holes (10 mm diame-ter by 15 mm deep) placed 15 mm apart in 3 rows of 3holes in a wooden base.18(p171) The maximum timeallowed to complete the test was 50 seconds, as mosthealthy adults complete the test in 18 seconds.18 Thescore was calculated as number of pegs per second.

b. Finger tapping. Patients were required to producerepetitive opposition of the index (first) finger to thethumb starting with digits together. The instructionwas to move the digits apart into an open position andthen back together as fast as possible while keeping asteady rhythm for 15 seconds. This task was undertakenfor both paretic and nonparetic hand, and a finger-tapping ratio was calculated (paretic:nonparetic).

5. Effect of MIT on quality of movement. The MotorActivity Log–Quality of Movement (MAL-QOM19)consists of a 14-item subjective assessment of how wellpatients perceive they have performed everyday func-tional activities using their paretic upper limb over theprevious week. Activities include brushing teeth,putting arm through clothing, and carrying an object.For each of the activities the patients report as per-forming in the previous week, they report how wellthey did it by using a 6-point scale from the pareticarm, which ranged from not used at all for that activity,score 0; to the ability to use the paretic arm for thatactivity was as good as before the stroke, score 5. Patientswere also able to select halfway scores, for example, 2.5.The MAL-QOM score was calculated as the meanscore for the activities reported as performed duringthe previous week.

6. Effect of MIT on functional ability. The Action ResearchArm Test (ARAT) measures functional use of theupper limb and is recognized as valid and reliable.20-22 Itis divided into 4 main sections of grasp, grip, pinch, andgross movement. Each section consists of 3, 4, or 6 itemsthat are scored between 0 and 3. The total possiblescore is 57.

Analysis

The proportion of patients who could performmotor imagery was calculated from the results of theCMIA. Information provided by patients on their per-ception of MIT was grouped thematically and used tomake appropriate adjustments to the content of MIT, tothe length of a session of MIT, and the timing of deliv-ery of MIT during a session. Thus, the content and dose

462

Tabl

e 1.

Ch

arac

teri

stic

s of

Incl

ude

d Pa

tien

ts a

nd

Th

ose

Perf

orm

ing

Ch

aoti

c M

otor

Im

ager

y

Mot

rici

tyTy

pe

ofSt

roke

Age

Tim

e A

fter

In

dex

– A

rm(B

amfo

rdPa

tien

t ID

(yea

rs)

Gen

der

Stro

ke (

mon

ths)

Pare

tic

Side

(bas

elin

e)cl

assi

fica

tion

)Le

sion

Loc

atio

n

Pati

ents

wh

o re

ceiv

ed M

IT20

0365

Mal

e18

Rig

ht

77LA

CI

Left

su

bcor

tica

l20

0982

Mal

e6.

2Le

ft92

PAC

IR

igh

t in

tern

al c

apsu

le20

1355

Fem

ale

30R

igh

t35

PAC

ILe

ft s

tria

to-c

apsu

lar

infa

rct

2017

74Fe

mal

e16

.5Le

ft77

LAC

IR

igh

t th

alam

ic20

2082

Mal

e14

Rig

ht

92TA

CI

Left

su

bcor

tica

l20

2449

Mal

e0.

5R

igh

t61

LAC

ILe

ft p

ons

Pati

ents

exc

lude

d fr

om M

IT b

ecau

seth

ey p

erfo

rmed

ch

aoti

c m

otor

imag

ery

2007

65Fe

mal

e0.

2Le

ft51

LAC

IR

igh

t in

tern

al c

apsu

le20

1874

Mal

e0.

9Le

ft92

LAC

IR

igh

t in

tern

al c

apsu

le20

2663

Mal

e0.

7Le

ft77

LAC

IR

igh

t in

tern

al c

apsu

le20

2960

Mal

e0.

1R

igh

t62

LAC

ILe

ft p

arav

entr

icu

lar

infa

rct

MIT

=m

otor

imag

ery

trai

nin

g;LA

CI

=la

cun

ar in

farc

tion

;PA

CI

=pa

rtia

l an

teri

or c

ircu

lati

on in

farc

tion

;TA

CI

=to

tal a

nte

rior

cir

cula

tion

infa

rcti

on.

Motor Imagery Poststroke: Dose and Aptness

Neurorehabilitation and Neural Repair 22(5); 2008 463

of MIT was changed in response to subject feedbackover the course of the study until no further changeswere suggested by patients. The change between base-line and outcome for motor and functional measureswas examined for the whole group and for each indi-vidual patient. Descriptive statistics were used toexplore group effects. Visual inspection of graphed datawas used to explore effects on individuals. Statisticaltesting was not undertaken because of the small samplesize and because randomization was not used for thisfeasibility study. Clinically important changes in motorscores were defined arbitrarily as:

• 10 points for the MI, as this is 10% of the total possiblescore

• 0.02 pegs per second for the 9HPT, as this represents 1peg placed in 50 seconds

• 0.1 for finger-tapping ratio, as this represents 10% oftotal possible score

• 0.5 for the MAL-QOM, as this represents 10% of thetotal possible score

• 5.7 points for the ARAT, as this represents 10% of thetotal possible score

RESULTS

Patients

Ten patients were included in this study of whom(40%) performed chaotic motor imagery (Table 1).Three patients showed evidence of temporal uncoupling,

and 1 patient performed executed movement duringmotor imagery (Table 2). Consequently, 6 patients(60%) were suitable and received MIT.

User Tolerability of Dose andContent of MIT

After the first MIT session, the patient reported dif-ficulty maintaining concentration after 20 minutes.Dose was therefore changed from two 30-minute periodsseparated by a 10-minute rest to two 20-minute periodsseparated by a 10-minute rest. All subsequent sessionsfor all patients were then delivered as two 20-minuteperiods separated by a 10-minute rest. Feedback fromall subsequent patient sessions was of difficulty in con-centrating as the 20-minute time point was reached.One patient found concentration dwindled due to thepredictable 3 × 3 structure of the intervention. Inresponse to complaints of lapses in concentration, thephysiotherapist used her judgment to request the per-formance of the selected action with varied speed,force, direction, or precision as the repetitionsadvanced (eg, knocking on table 3 times, repeat withslightly increased force, repeat with further increasedforce). No negative feedback was received about theactivities of MIT.

Subject Opinion

All opinions and comments received from patientswere positive and informative. No detrimental effectswere reported. In summary, patients reported increasedfunctional use and quality of movement of the pareticupper limb. They also reported becoming more aware ofthe paretic upper limb and resuming exercises.

Effect on Muscle Weakness

Between baseline and outcome, 4 of the 6 patientsincreased MI score (minimum increase of 8 and maxi-mum increase of 16 points) between baseline and out-come, whereas 1 patient had a decreased score and 1patient showed no change (Table 3 and Figure 1). Thus, 3of the 6 patients showed clinically important improvementin muscle weakness, although it must be appreciatedthat the patient who increased motricity score by 8points reached a maximum score. Of those patients whoimproved on the MI, only 1 deteriorated between out-come and follow-up, whereas 3 maintained their gains(Table 3 and Figure 1).

Table 2. Chaotic Motor Imagery Assessment of All Patients

Hand MotorRotation Imagery Break Point

Patient ID % correct Finger Tapping MI-EM %

Able to performmotor imagery

2003 87 Correct –272009 93 Correct –12013 95 Correct –422017 93 Correct –152020 77 Correct –192024 99 Correct –24

Found to performchaotic motorimagery

2007a 88 Correct XS2018a 90 Correct +322026a 89 Correct +172029a 99 Correct +12

aExcluded due to incompatible break point or XS (executed movement).

464

Tabl

e 3.

Scor

es f

or M

otri

city

In

dex–

Arm

,Nin

e-H

ole

Peg

Test

,Fin

ger

Tapp

ing,

and

Mot

or A

ctiv

ity

Log

at B

asel

ine,

Ou

tcom

e,an

d Fo

llow

-Up

Mea

n (

SD)

Mea

n (

SD)

Mea

n (

SD)

Pati

ent

IDB

asel

ine

[Med

ian

(IQ

R)]

Ou

tcom

e[M

edia

n (

IQR

)]Fo

llow

-Up

[Med

ian

(IQ

R)]

Mot

rici

ty I

nde

x–A

rm20

0377

72.3

3 (2

1.63

)77

78.8

3 (1

7.10

)77

76.6

7 (1

5.88

)20

0992

[77.

00 (

37.5

0)]

100

[77.

00 (

23.7

5)]

100

[77.

00 (

14.0

0)]

2013

3550

5020

1777

9279

2020

9277

7720

2461

7777

Nin

e-H

ole

Peg

Test

(no

ofpe

gs/s

)20

030.

080.

17 (

0.15

)0.

120.

17 (

0.14

)0.

130.

18 (

0.11

)20

090.

34[0

.18

(0.3

0)]

0.36

[0.1

8 (0

.27)

]0.

34[0

.20

(0.1

6)]

2013

00

020

170.

30.

280.

2320

200.

280.

230.

220

240.

020.

040.

19Fi

nge

r-ta

ppin

g ra

tio

pare

tic:

non

pare

tic

2003

0.49

0.42

(2.

95)

0.5

0.46

(0.

28)

0.55

0.53

(0.

31)

2009

0.66

[0.5

3 (0

.59)

]0.

65[0

.57

(0.5

0)]

0.72

[0.6

2 (0

.48)

]20

130

00

2017

0.69

0.67

0.87

2020

0.56

0.7

0.69

2024

0.1

0.24

0.37

Mot

or A

ctiv

ity

Log

2003

1.12

1.60

(1.

19)

1.75

2.28

(1.

32)

1.64

2.01

(1.

26)

2009

2.25

[1.6

9 (2

.31)

]2.

88[2

.32

(2.4

2)]

1.2

[1.4

8 (2

.30)

]20

130.

150.

460.

7520

172.

573.

863.

9320

203.

043.

423.

220

240.

461.

331.

31A

ctio

n R

esea

rch

Arm

Tes

t20

0348

39.8

3 (2

1.33

)50

43.1

7 (1

9.50

)51

46.6

7 (2

0.54

)20

0957

[48.

50 (

38.0

0)]

57[5

1.50

(29

.00)

]57

[0.5

5 (1

76.5

0)20

134

75

2017

5757

5720

2049

5356

2024

2435

54

Motor Imagery Poststroke: Dose and Aptness

Neurorehabilitation and Neural Repair 22(5); 2008 465

Effect on Hand Dexterity

1. 9HPT. Between baseline and outcome, 3 patientsshowed improvement in the 9HPT from 0.02 to 0.04

pegs per second (Table 3 and Figure 2). Another hadno change, and 2 deteriorated (Table 3 and Figure 2).Between outcome and follow-up, 1 of the patients whoimproved at outcome showed a substantial furtherimprovement of 0.15 pegs per second, 1 a modest fur-ther improvement of 0.02 pegs per second, whereas thethird patient’s ability returned to the baseline level(Table 3 and Figure 2).

2. Finger tapping. Between baseline and outcome, 3patients improved finger-tapping ratio by 0.14 (Table3 and Figure 3). Of the remaining 3 patients, 1showed no change and 2 deteriorated (Table 3 andFigure 3).

Effect on Quality of Movement

Between baseline and outcome, all 6 patients hadimproved quality of movement scores (Table 3 andFigure 4).

Effect on Functional Ability

At baseline, 2 patients obtained maximal scores andmaintained this at outcome and follow-up, so they areomitted from these change results. Between baselineand outcome, the 4 patients, who had not reached theceiling for the ARAT, improved minimally (Table 3 andFigure 5).

DISCUSSION

This feasibility study found that MIT was receivedpositively by patients and that the tolerable daily dosewas 60 minutes split into two 20-minute sessions of

30

40

50

60

70

80

90

100

Baseline Outcome Follow-up

2003

2009

2013

2017

2020

2024

Figure 1. Motricity Index scores at baseline, outcome, andfollow-up for all patients who were able to perform motorimagery.

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

0.55

0.6

0.65

0.7

Baseline Outcome Follow-up

2003

2009

2013

2017

2020

2024

Figure 2. Nine-Hole Peg Test scores (number of pegs persecond) at baseline, outcome, and follow-up for all patientswho were able to perform motor imagery.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Baseline Outcome Follow-up

2003

2009

2013

2017

2020

2024

Figure 3. Finger-tapping ratio, paretic:nonparetic, at base-line, outcome, and follow-up for all patients who were able toperform motor imagery.

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

Baseline Outcome Follow-up

2003

2009

2013

2017

2020

2024

Figure 4. Motor Activity Log–Quality of Movement at base-line, outcome, and follow-up for all patients who were able toperform motor imagery.

Simmons et al

466 Neurorehabilitation and Neural Repair 22(5); 2008

MIT separated by a 10-minute rest. This daily dose differsfrom those used in earlier controlled studies, which pro-vided 10 minutes,23 30 minutes,24 1 hour,25 and inde-pendent practice with no duration specified.26 There is,however, no indication in these earlier studies thatpatients were asked about tolerability of the daily dose.This present study, which adjusted daily dose to con-sider reported attention span of patients, has thereforeprovided a starting point for the process of dose-findingfor MIT for people after subcortical stroke.

The content of the MIT was acceptable to patients,but we found that the mode of delivery needs to bevaried in terms of instructions for different speeds,force, direction, or precision. No adverse effects werereported by patients, and none of the patients withdrewfrom MIT. Exploration of the effects of MIT on aspectsof motor function and functional ability revealedinterindividual variation in changes (outcome minusbaseline) in motor function and functional ability, butno clinically important detrimental changes occurredexcept for the 9HPT score for 1 patient (patient 2020).In summary, we found possibly beneficial changes atoutcome in muscle strength (3 of 6 patients), hand dex-terity (3 of 6 patients), finger tapping (2 of 6 patientswith better quality of movement in all 6), and func-tional ability (1 of 4 patients).

In this study, 40% of patients were excluded becauseof performance of chaotic motor imagery. Interestingly,those patients excluded do not appear to differ in termsof age, lesion location, medication, or risk factors, buttime from onset of stroke may predispose to impairedperformance. Previous behavioral studies have excludedpatients with parietal lobe involvement, but our datasuggest that screening based on lesion location27,28

might be insufficient. Thus, a more rigorous approachmay be needed in future studies to exclude patients per-forming chaotic motor imagery, a potential confounderof prior studies.23-26,29,30 However, further characterization

of those patients performing chaotic motor imagery isneeded to assess whether they are able to “recover” theability to perform motor imagery or whether they havespecific deficits—stroke-related or not—in motor per-formance. Moreover, it is possible that patients whoperform chaotic motor imagery can gain from MIT.Further work is required to investigate whether patientswho do not pass the chaotic motor imagery screening,as well as those with high and low levels of motor con-trol, can benefit from MIT.

ACKNOWLEDGMENTS

Work reported in this article is supported by TheStroke Association (TSA 2003/10) and the MedicalResearch Council (MRC G0001219). NS is supported bya Brain Entry Scholarship, The Stroke Association (TSA2003/10) and Sackler Fellowship. LS is supported byThe Stroke Association (TSA 2003/10).

APPENDIX 1.

Motor Imagery Assessment

The combination of these 3 assessments has beentermed the Chaotic Motor Imagery Assessment(CMIA), and the sequence is crucial in preserving theunderlying cognitive assumptions.10 During all tasksrequiring explicit motor imagery, patients were givenspecific instructions to perform first-person motorimagery, not to view the scene from the third person,and not to count or assign numbers or tones to each fin-ger. The complete assessment is detailed elsewhere.5

Assessment 1: Hand RotationPatients were presented with 96 picture cards of

hands—4 different views, 12 rotations (30 degree steps),left & right14—with their hands palm down on their lap,and asked to respond whether the presented picture wasof a left or right hand. This task is widely accepted toactivate the motor imagery network.15 Patients wereexcluded if they scored below 75% (chance level 50%).

Assessment 2: Simple Failure to ComplyThe premise behind this next assessment is to test

whether the patient is able to actively engage and main-tain the motor imagery network in a task. The patientswere taught an auditory paced (1 Hz) finger-thumbopposition sequence (2,3,4,5,2 . . .) using executedmovement, which alternated with periods of rest in ablock design. Once executed performance was perfect, itwas repeated using motor imagery. To monitor compli-ance, the block length was varied pseudo-randomly and

0369

12151821242730333639424548515457

Baseline Outcome Follow-up

2003

2009

2013

2017

2020

2024

Figure 5. Action Research Arm Test at baseline, outcome, andfollow-up for all patients who were able to perform motor imagery.

Motor Imagery Poststroke: Dose and Aptness

Neurorehabilitation and Neural Repair 22(5); 2008 467

the patient had to confirm his or her position within thesequence after each block by touching the appropriate“stop” finger.

Assessment 3: Alternative Cognitive StrategiesA priori it was anticipated that most patients would

correctly perform Assessment 2 but that some may usealternative cognitive strategies such as counting fingers.Therefore, a simple task, based on Fitt’s law (ie, speed/accuracy trade-off), was adapted16 to test and excludesuch patients.

In a controlled environment, patients performed thefinger-tapping sequence (2,3,4,5,2 . . .), initially usingexecuted movement and then motor imagery; however,in contrast to Assessment 2, the external auditory pac-ing rate was gradually increased, initially starting at 40beats/min and increasing by 10 beats/min every 5 sec-onds. The “break point,” defined as when the patientwas no longer able to perform the task accurately, wasmeasured. Patients were excluded if the mean breakpoint was greater for motor imagery compared to exe-cuted movement. No attempts were made to eitheradjust or explore the cognitive strategy employed insuch cases other than simple debriefing. In strokepatients, when unable to perform the executed task, andthus to provide the reference break point with theaffected hand due to poor recovery, the nonaffectedhand was used.

REFERENCES

1. Kwakkel G, Kollen B, Wagenaar RC. Long term effects of inten-sity of upper and lower limb training after stroke: a randomisedtrial. J Neurol Neurosurg Psychiatry. 2002;72:473-479.

2. Dobkin B. Confounders in rehabilitation trials of task-orientedtraining: lessons from the designs of the EXCITE and SCILTmulticenter trials. Neurorehabil Neural Repair. 2007;21:3-13.

3. Kimberley T, Khandekar G, Skraba L, et al. Neural substrates formotor imagery in severe hemiparesis. Neurorehabil NeuralRepair. 2006;20:268-277.

4. Lotz M, Braun C, Birbaumer N, et al. Motor learning elicited byvoluntary drive. Brain. 2003;126:866-872.

5. Sharma N, Simmons LH, Jones PS, et al. The neural substrates ofmotor imagery in well recovered sub-cortical stroke patients.Paper submitted.

6. Bonnet M, Decety J, Jeannerod M, et al. Mental simulation of anaction modulates the excitability of spinal reflex pathways inman. Cogn Brain Res. 1997;5:221-228.

7. Li S, Kamper DG, Stevens JA, et al. The effect of motor imageryon spinal segmental excitability. J Neurosci. 2004;24:9674-9680.

8. Jeannerod M. Mental imagery in the motor context. Neuropsy cholo-gia. 1995;33:1419-1432.

9. Berthoz A. The role of inhibition in the hierarchical gating of exe-cuted and imagined movements. Cogn Brain Res. 1996;3:101-113.

10. Sharma N, Pomeroy VM, Baron J-C. Motor imagery. A backdoorto the motor system after stroke. Stroke. 2006;37:1941-1952.

11. Warner L, McNeill ME. Mental imagery and its potential forphysical therapy. Phys Ther. 1988;4:516-521.

12. Pomeroy VM, Cooke E, Hamilton S, et al. Development of a sched-ule of current physiotherapy treatment used to improve movementcontrol and functional use of the lower limb after stroke: a precur-sor to a clinical trial. Neurorehabil Neural Repair. 2005;19:350-358.

13. Oldfield R. The assessment and analysis of handedness: theEdinburgh inventory. Neuropsychologia. 1971;9:97-113.

14. Nico D, Daprati E, Rigal F, et al. Left and right hand recognitionin upper limb amputees. Brain. 2004;127:120-132.

15. Parsons M, Fox PT, Downs JH, et al. Use of implicit motorimagery for visual shape discrimination as revealed by PET.Nature. 1995;375:54-58.

16. Sirgu A, Duhamel JR, Cohen L, et al. The mental representationof hand movements after parietal cortex damage. Science.1996;273:1564-1568.

17. Demeurisse G, Demol O, Robaye E. Motor evaluation in vascularhemiplegia. Eur Neurol. 1980;19:382-389.

18. Wade DT. 1992. Measurement in Neurological Rehabilitation.Oxford, UK: Oxford University Press; 1992.

19. Uswatte G, Taub E, Morris D, et al. Reliability and validity of theupper-extremity motor activity log-14 for measuring real-worldarm use. Stroke. 2005;36:2493-2496.

20. De Weerdt WJG, Harrison MA. Measuring recovery of arm-handfunction in stroke patients: a comparison of the Brunnstrom-Fugl-Meyer test and the Action Research Arm Test. PhysiotherCan. 1985;27:65-70.

21. Hsieh CL, Hsueh IP, Chiang FM, et al. Inter-rater reliability andvalidity of the Action Research Arm Test in stroke patients. AgeAgeing. 1998;27:107-113.

22. Van der Lee JH, De Groot V, Beckerman H, et al. The intra- andinterrater reliability of the Action Research Arm Test: a practicaltest of upper extremity function in patients with stroke. ArchPhys Med Rehabil. 2001;82:14-19.

23. Page SJ, Levine P, Sisto S, et al. A randomized efficacy and fea-sibility study of imagery in acute stroke. Clin Rehabil.2001;15:233-240.

24. Page SJ, Levine P, Leonard AC. Effects of mental practice onaffected limb use and function in chronic stroke. Arch Phys MedRehabil. 2005;86:399-402.

25. Dijkerman HC, Johnston LM, MacWalter RS. Does motorimagery training improve hand function in chronic strokepatients? A pilot study. Clin Rehabil. 2004;18:538-549.

26. Dijkerman HC, Johnston LM, MacWalter RS. Does motorimagery training improve hand function in chronic strokepatients? A pilot study. Clin Rehabil. 2004;18:538-549.

27. Johnson SH. Imagining the impossible: intact motor representa-tions in hemiplegics. Neuroreport. 2000;11:729-732.

28. Johnson SH, Sprehen G, Saykin AJ. Intact motor imagery inchronic upper limb hemiplegics: evidence for activity-indepen-dent action representations. J Cogn Neurosci. 2002;14:841-852.

29. Liu KP, Chan CC, Lee TM, et al. Mental imagery for promotingrelearning for people after stroke: a randomized controlled trial.Arch Phys Med Rehabil. 2004;85:1403-1408.

30. Dunsky A, Dickstein R, Ariav C, et al. Motor imagery practice ingait rehabilitation of chronic post-stroke hemiparesis: four casestudies. Int J Rehabil Res. 2006;29:351-356.