G Model

REHAB 1192 1–9

Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx

Available online at

ScienceDirect
www.sciencedirect.com

1
2 Original article

3 Descriptive pilot study of vividness and temporal equivalence

4 during motor imagery training after quadriplegia
5 Q1 Sébastien Mateo a,b,c,d,*, Karen T. Reilly a, Christian Collet d, Gilles Rode a,b
6 a
Inserm U1028, CNRS UMR5292, Lyon Neuroscience Research Center, ImpAct Team, Université de Lyon, Université Lyon 1, 69676 Lyon, France
7 b
Plate-forme Mouvement et Handicap, Hospices Civils de Lyon, hôpital Henry Gabrielle, 69000 Lyon, France
8 c
École Normale Supérieure de Lyon, CNRS UMR5672, Université de Lyon, Université Lyon 1, 69007 Lyon, France
9 d
Laboratoire interuniversitaire de la biologie de la motricité LIBM, équipe d’Accueil 7424, Université de Lyon, Université Lyon 1, 69622 Villeurbanne cedex,
10 France

A R T I C L E I N F O A B S T R A C T

Article history: Background: Motor imagery (MI) training is often used to improve physical practice (PP), and the
Received 16 October 2017 functional equivalence between imagined and practiced movements is widely considered essential for
Accepted 2 June 2018 positive training outcomes.
Objective: We previously showed that a 5-week MI training program improved tenodesis grasp in
Keywords: individuals with C6–C7 quadriplegia. Here we investigated whether functional equivalence changed
Mental practice during the course of this training program.
Tetraplegia
Methods: In this descriptive pilot study, we retrospectively analyzed data for 6 individuals with C6–C7
Tenodesis grasp
Rehabilitation
quadriplegia (spinal cord injured [SCI]) and 6 healthy age-matched controls who trained for 5 weeks in
Vividness visual and kinesthetic motor imagery or visualization of geometric shapes (controls). Before training, we
Mental chronometry assessed MI ability by using the Kinesthetic and Visual Imagery Questionnaire (KVIQ). We analyzed
functional equivalence by vividness measured on a visual analog scale (0–100) and MI/PP time ratios
computed from imagined and physically practiced movement durations measured during MI training.
These analyses were re-run considering that half of the participants with quadriplegia were good
imagers and the other half were poor imagers based on KVIQ scores. To investigate generalization of
training effects, we analyzed MI/PP ratios for an untrained pointing task before (3 baseline measures),
immediately after, and 2 months after training.
Results: During MI training, imagery vividness increased significantly. Only the good imagers evolved
toward temporal equivalence during training. Good imagers were also the only participants who showed
changes in temporal equivalence on the untrained pointing task.
Conclusion: This is the first study reporting improvement in functional equivalence during an MI training
program that improved tenodesis grasp in individuals with C6–C7 quadriplegia. We recommend that
clinical MI programs focus primarily on vividness and suggest that feedback about movement duration
could potentially improve temporal equivalence, which could in turn lead to further improvement in PP.
C 2018 Elsevier Masson SAS. All rights reserved.

11
12 1. Introduction preventing effective grasping movements [2]. Grasping is a 18
primary rehabilitation aim [3] because regaining functional 19
13 Quadriplegia is a severe condition consecutive to a cervical grasping greatly enhances independence [4]. Despite their injury, 20
14 spinal cord injury (SCI) sparing the brain [1]. Depending on the individuals with C6–C7 quadriplegia can learn to perform a passive 21
15 level of injury, the resulting de-efferentation reduces the number grip (either lateral or palmar) by using the tenodesis grasp in which 22
16 of degrees of freedom, whereas de-afferentation impairs somato- wrist extension leads to flexor tendon shortening [5,6]. 23
17 sensory and haptic inputs from the upper limb, both altering or Mental rehearsal of upper-limb movements seems to be a 24
promising adjunct therapy to classical grasping rehabilitation 25
programs [7,8]. Motor imagery (MI) is the construction of a mental 26
* Corresponding author. Hôpital Henry Gabrielle, 20, route de Vourles, 69230
representation from an internal perspective based on visual or 27
Saint-Genis Laval, France. kinesthetic information without simultaneous physical execution 28
E-mail address: sebastien.mateo@chu-lyon.fr (S. Mateo). [9]. MI involves motor programming similar to that with physical 29

https://doi.org/10.1016/j.rehab.2018.06.003
1877-0657/ C 2018 Elsevier Masson SAS. All rights reserved.

Please cite this article in press as: Mateo S, et al. Descriptive pilot study of vividness and temporal equivalence during motor imagery
training after quadriplegia. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.003
 G Model
REHAB 1192 1–9

2 S. Mateo et al. / Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx

30 practice (PP) but with inhibition of motor output at some level of temporal equivalence was measured while participants performed 92
31 the corticospinal flow [10]. Both PP and MI activate common brain a complex pointing movement (untrained movement); 5 weeks of 93
32 regions [11], which suggest the engagement of similar internal MI training (weeks 6 to 10, 3 sessions per week, each lasting 94
33 models of action [12], and show similar movement durations (i.e., 45 min) during which temporal equivalence and imagery vividness 95
34 temporal equivalence [13]), which reflects the sharing of neural were measured for a single joint movement and reach-to-grasp 96
35 processes [14]. Activity in common neural circuits is thought to movements (trained movements); and a post-training phase in 97
36 underlie improvements in motor performance observed after MI which temporal equivalence was measured for the untrained 98
37 training in both healthy individuals [15] and those with SCI [7,8]. pointing movement immediately after training and 8 weeks later 99
38 MI training programs aim to achieve functional equivalence: a (weeks 11 and 19, respectively). Pre- and post-training phases 100
39 vividness and duration of imagined movements as close as possible were identical for control participants and those with quadriplegia. 101
40 to that of physically practiced movements [14]. MI vividness refers However, 5 weeks of training in controls consisted of visualizing 102
41 to the clarity, brightness, or intensity of the mental reconstruction geometric shapes (3 sessions per week, each lasting 45 min) during 103
42 of the movement [16]. It is a marker of MI quality and is usually which no measures of imagery performance were taken. 104
43 reported by participants on a visual analog scale (VAS) or During the 18 weeks of the experiment, participants with C6– 105
44 questionnaires designed for specific populations [e.g., for individ- C7 quadriplegia continued their usual rehabilitation program, 106
45 uals with SCI, the Kinesthetic and Visual Imagery Questionnaire which consisted of twice daily 30-min sessions of physical therapy 107
46 (KVIQ, designed for people with motor impairments [17])]. and daily 30-min session of occupational therapy, every weekday. 108
47 Previous reports in healthy individuals have found high vividness This program consisted of strengthening and task-oriented 109
48 scores related to greater performance improvement [18,19]. This practice including reach-to-grasp toward a range of objects with 110
49 result emphasizes the interest of using MI vividness as a marker of varied locations, orientations and sizes. 111
50 MI quality [20]. The temporal equivalence between MI and PP is
51 also used to assess MI quality [21], and studies of healthy 2.2. Population 112
52 individuals and those with neurological impairments (including
53 those with SCI) show that PP and MI have similar durations but Participants were 6 inpatients with chronic C6–C7 SCI [mean 113
54 lower vividness on the affected side [22,23]. In healthy individuals (SD) age 30.3 (8.2), ASIA Impairment Scale A–B] and 6 healthy 114
55 (athletes), MI training results in greater performance improvement control participants matched on age, sex, and handedness [mean 115
56 in good imagers [15,18,19,25]. However, in individuals with (SD) age 28.3 (6.9)] [7]. Because extensor muscle strength reaches a 116
57 quadriplegia, little is known about the functional equivalence plateau at about 6 months post-injury [26], we included only 117
58 between MI and PP, whether it can be modified by MI training, and individuals at least 6 months after their injury. Five participants 118
59 if so, whether there is a relation between MI training effects and with quadriplegia were right-handed and one was left-handed 119
60 initial imagery capacity. according to the Edinburgh questionnaire [27]. All participants had 120
61 In a previous study of 6 individuals with C6–C7 quadriplegia, 1) no history of traumatic brain injury associated with the SCI, 2) 121
62 we showed that 5 weeks of MI training improved tenodesis grasp normal or corrected-to-normal vision, 3) normal control of the 122
63 and induced plastic changes in the sensorimotor cortex [7]. These wrist extensor muscles (manual muscle test = 5/5 according to 123
64 results suggest the potential interest of including MI in hand international standards [1]), 4) complete paralysis of finger flexors 124
65 rehabilitation after quadriplegia, but several questions remain to (0/5), 5) normal joint motion of the upper limb as measured by 125
66 be answered before investing in a large-scale study designed goniometry, and 6) no spasticity of the upper limb muscles (i.e., 0/4 126
67 specifically to test the efficacy of MI training to improve grasping. on the Ashworth modified scale [28]). For each participant, we 127
68 Here, we present a descriptive pilot study in which we administered the KVIQ once during pre-training. We then averaged 128
69 retrospectively analyzed MI data collected (but not included) in the scores for the 20 items of the KVIQ, and using a cut-off of 3.75/5 129
70 the above-mentioned kinematic/magnetoencephalography study. (proposed by McInnes et al. for stroke [29]), half of the participants 130
71 The aim of the current study was to ask whether MI training in with quadriplegia were classified as having good MI ability and the 131
72 individuals with quadriplegia alters functional equivalence, other half poor MI ability. Characteristics of participants with 132
73 whether this depends on imagery ability, and whether MI training quadriplegia are summarized in Table 1. 133
74 generalizes to a non-trained movement.
2.3. Measure of temporal equivalence during a non-trained task 134

75 2. Methods All participants imagined and physically practiced a pointing 135

task 3 times before and 2 times after training. For this task, we 136
76 2.1. Study design and setting positioned an A4 sheet flat on the table displaying 2 circles 137
(diameters 30 and 84 mm, respectively; line thickness 1 mm) 138
77 Details of the initial longitudinal case-series study in which the placed one inside the other with a 2-mm diameter solid circle 139
78 data presented here were collected have been published elsewhere placed at their center. Each circle had a mark at 3, 6, 9, and 140
79 [7]. Briefly, the aim of the initial study was to examine tenodesis 12 o’clock (marker length 12 mm, line thickness 1 mm). All 141
80 kinematics and brain plasticity after 5 weeks of MI training. It was participants pointed with the index finger of their dominant arm. 142
81 part of a Hospital Clinical Research Program (HCRP-2008-541/142; Starting from the center, participants touched each of the 4 marks 143
82 ClinicalTrials.gov identifier NCT02860403). After gaining approval on the small circle starting from 12 o’clock and turning clockwise. 144
83 from the ethics committee of the university hospital (CPP 2009- After each mark had been touched, they had to return to the center 145
84 051 B), all participants provided written informed consent in point (e.g., center–12–center–3–center–6–center–9). The partici- 146
85 accordance with the Declaration of Helsinki. pant then traced the circle clockwise with their index finger 147
86 The current descriptive pilot study retrospectively analyzed the starting from the center then moving up to 12 and then traced the 148
87 effect of this MI training protocol on imagery vividness and the entire circle before returning to the center. The participant then 149
88 temporal equivalence (ratio of MI to PP duration) of the trained repeated the same pointing and tracing sequence for the larger 150
89 movements and on a complex, untrained pointing movement. Data circle. We measured the total time for both imagined and 151
90 were collected over 18 weeks in 3 different phases: a pre-training physically practiced movements in order to compare their 152
91 phase (weeks 1 to 5, before MI training began) during which durations and to test temporal equivalence. For imagined 153

Please cite this article in press as: Mateo S, et al. Descriptive pilot study of vividness and temporal equivalence during motor imagery
training after quadriplegia. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.003
 G Model
REHAB 1192 1–9

S. Mateo et al. / Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx 3

Table 1 and were designed to train participants to imagine movements 165

Q5 Characteristics of the 6 participants with quadriplegia.
important for tenodesis grasp. We thus trained their dominant 166
Participants upper limb on a single-joint wrist extension movement and multi- 167
P1 P2 P3 P4 P5 P6 joint palmar and lateral grip prehension movements. All move- 168
ments were performed at a comfortable, self-paced speed, and we 169
Sex Male Female Male Male Female Male
Hand dominancea R R R L R R
assessed both MI vividness and movement duration. 170
Age (years) 18 31 24 37 32 40 Single- and multi-joint movements were performed in each MI 171
Months post-injury 14 30 6 13 6 13 session but in separate blocks, for a total of 6 to 8 blocks in each 172
ASIA level (dom.) C7 C6 C6 C6 C6 C6 session. Thus, in each session, participants performed 3 or 4 of each 173
ASIA level (non-dom.) C6 C6 C7 C5 C7 C5
movement type. A block consisted of 4 sequential events: 174
AIS score (A–E) A A B A B B
UEMS (/50) 26 24 26 15 27 17 PP of the movement once; 175
UEMS dom. (/25) 15 12 11 9 11 10 from 5 to 9 repetitions of visual MI (VMI); 176
UEMS non-dom. (/25) 11 12 15 6 16 7 PP of the movement once, and; 177
Sensitive ASIA score (/112) 50 28 68 27 48 56 from 5 to 9 repetitions of kinesthetic MI (KMI). 178
KVIQ-20 kinesthetic scoreb 4.9 4.8 2.8 3.2 3.2 3.6
The experimenter described the movement to be performed by 179
KVIQ-20 visual scoreb 4.7 3.9 2.4 2.9 3.0 4.4
KVIQ-20 total scoreb 4.8 4.4 2.6 3.1 3.1 4.0 reading a script (using a first-person perspective; see Appendix) 180
and announced the number of repetitions to perform. In line with 181
R: right; L: left; ASIA: American Spinal Injury Association; AIS: ASIA Impairment
Scale; UEMS: Upper Extremity Motor Score; dom: dominant; non-dom: non- the clinical objective of training, the tenodesis grasp, and to 182
dominant; KVIQ: Kinesthetic and Visual Imagery Questionnaire. potentiate the generalization of MI training to complex, multi-joint 183
a
After completion of the Edinburgh questionnaire. movements, we changed between lateral and palmar tenodesis 184
b
KVIQ-20 results are expressed as mean values, where the mean is the average of grips for the first 4 weeks of training and trained both grips in the 185
10 imagined movements (for the kinesthetic and visual sub-scores) or of all
final training week. All participants were instructed to avoid 186
20 movements (for the total score). KVIQ vividness scores ranged from 1 to 5, with
1 corresponding to the absence of an image or perception/feeling, and 2 to poor, additional mental rehearsal outside MI sessions and were 187
3 moderate, 4 good, and 5 very good image or perception/feeling. Mean (SD) scores interviewed before each session to ensure compliance with this 188
on the KVIQ were 4.0 (0.3) (total), 4.3 (0.4) (VMI), and 3.7 (0.4) (KMI) for control instruction. 189
participants 3.7 (0.9) (total), 3.6 (0.9) (VMI), and 3.8 (0.9) (KMI) for participants with
After every single imagined movement (VMI and KMI), 190
quadriplegia.
participants rated perceived movement vividness on a VAS (0– 191
100 mm), 0 representing failure to imagine the movement and 192
154 movements, participants verbally indicated the start and stop 100 representing highly vivid MI (i.e., image as clear or perception 193
155 times, but for physically practiced movements the physical as intense as during PP). The duration of every single imagined and 194
156 therapist gave the go signal and stopped the stopwatch when physically practiced movement was measured by using a 195
157 the movement was completed. The instructions given to partici- stopwatch. For PP movements, the experimenter verbally triggered 196
158 pants about the imagery were to imagine the same movement as the start time and visually observed movement completion. For 197
159 they had just physically executed at the same speed. imagined movements, participants verbally indicated the start and 198
stop times. 199
160 2.4. MI training and functional equivalence
2.5. Data processing 200
161 Participants with quadriplegia underwent MI training while
162 Q2 continuing their usual rehabilitation program (Fig. 1). Each MI For each participant, the duration of each imagined movement 201
163 session lasted 45-min and sessions were repeated 3 times a week, was converted into a ratio by dividing the MI duration by the PP 202
164 over 5 weeks. MI sessions were supervised by a physical therapist duration. For example, for a wrist extension block, the duration of 203

Fig. 1. Description of the study protocol. Temporal equivalence on a complex pointing task was assessed 3 times before training and 2 times after (short arrows) and the
Kinesthetic and Visual Imagery Questionnaire (KVIQ) was assessed once before training (asterisks in the figure). Motor imagery (MI) training consisted of 6 to 8 blocks of MI
during a 45-min session (long arrows) that was repeated 3 times a week for 5 weeks. A single block consisted of 1) one single physically practiced (PP) movement and several
repetitions of visual MI (VMI) [n = 6 (week 1) to 9 (week 5)] and 2) one single PP movement, then several repetitions of kinesthetic MI (KMI) [n = 6 (week 1) to 9 (week 5)]. PT:
physical therapy; OT: occupational therapy; LG: lateral grip; PG: palmar grip; W: week.

Please cite this article in press as: Mateo S, et al. Descriptive pilot study of vividness and temporal equivalence during motor imagery
training after quadriplegia. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.003
 G Model
REHAB 1192 1–9

4 S. Mateo et al. / Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx

204 each of the VMI movements in that block was divided by the 3.2. Trained movements: changes in vividness and temporal 262
205 duration of the single PP movement at the start of the block. We equivalence 263
206 repeated the same process for KMI. Hence, we calculated a mean
207 MI/PP ratio separately for visual and kinesthetic imagery The evolution of VMI and KMI vividness scores during training 264
208 repetitions for each MI training block. Because each session (weeks 6 to 10) is shown in the top row of Figs. 2 and 3. The LMM 265
209 contained several blocks of the same movement, we averaged revealed a significant change in VMI and KMI vividness for both 266
210 across blocks so that for each session we had a single ratio for each single-joint (Fig. 2) (x2(4) = 26.0, P < 0.001 and x2(4) = 49.9, 267
211 movement (single- versus multi-joint) and for each imagery P < 0.001) and multi-joint (Fig. 3) movements (x2(4) = 22.2, 268
212 modality (visual versus kinesthetic). P < 0.001 and x2(4) = 41.5, P < 0.001). All four post-hoc tests 269
comparing week 6 to week 10 had P-values < 0.01, indicating that 270
vividness increased significantly for both imagery modalities and 271
213 2.6. Statistical analyses
both movement types. For single-joint movements, vividness 272
scores increased by 21% and 31% (VMI and KMI), and for multi-joint 273
214 For the MI training data, we investigated whether the
movements scores by 16% and 32% (VMI and KMI). 274
215 vividness scores and MI/PP ratios changed over the 5 weeks of
The bottom row of Figs. 2 and 3 shows the evolution of VMI/PP 275
216 MI training. We analyzed data for VMI and KMI as well as single-
and KMI/PP time ratios across training. The MI/PP time ratio 276
217 and multiple-joint movements separately. We entered data from
(temporal equivalence) did not change for VMI or KMI during 277
218 each session into a Linear Mixed Model (LMM) considering
single-joint movements (x2(4) = 3.85, P = 0.43, and x2(4) = 2.91, 278
219 changes across weeks as fixed effects and participants as random
P = 0.57) or during multi-joint movements for KMI (x2(4) = 6.89, 279
220 effects, thus integrating individual differences into the model.
P = 0.14). In contrast, temporal equivalence evolved significantly 280
221 For exploratory purposes, we repeated the LMM twice, once for
during VMI of multi-joint movements (x2(4) = 15.4, P = 0.004), with 281
222 poor imagers and once for good imagers. For all analyses, we
VMI/PP time ratios decreasing by 22% between week 6 and 10 282
223 performed an additional LMM to verify the absence of changes
(post-hoc test P = 0.02) and ending at a ratio close to 1. 283
224 within each week, considering each session within a week as a
The above analyses were re-run separately for good and poor 284
225 fixed effect.
imagers. During single-joint movements, good imagers showed no 285
226 To test whether the training intervention had an effect on
significant changes in vividness for VMI but significant changes for 286
227 temporal equivalence measured during the untrained pointing 2 2
KMI (x (4) = 8.8, P = 0.07, x (4) = 32.2, P < 0.001). During multi-joint 287
228 task, we also applied an LMM to the pointing-task MI/PP ratios,
movements, both VMI and KMI changed significantly (x2(4) = 18.2, 288
229 separately for individuals with quadriplegia and control partici-
P = 0.001, x2(4) = 18.7, P < 0.001). Of the 3 post-hoc tests compar- 289
230 pants; the model compared the 3 pre-training tests to the 2 post-
ing week 6 to week 10, only the KMI comparisons were significant 290
231 training tests with participants as random effects. To verify the
(both P < 0.05 VMI multi-joint; P = 0.07) and KMI vividness 291
232 absence of changes between the 3 pre-training tests, we also
increased across training by 31% and 26% (single- and multi-joint 292
233 performed additional LMMs on these data.
movements). Importantly, at week 10, all vividness values for good 293
234 For all models, we ensured compliance with LMM hypotheses
imagers were within the range considered normal (blue zone on 294
235 by testing that the residuals were normally distributed by using
the figures). MI/PP time ratios significantly changed across training 295
236 the Kolmogorov–Smirnov test. If the LMM reached the signifi-
for both VMI and KMI for multiple-joint movements (x2(4) = 19.5, 296
237 cance threshold (P < 0.05), we performed post-hoc analyses by
P < 0.001, x2(4) = 18.3, P = 0.001) but not single-joint movements 297
238 using least mean squares with Tukey correction (LS post-hoc).
(x2(4) = 8.2, P = 0.08, x2(4) = 6.4, P = 0.2). P-values for the 2 post-hoc 298
239 When data showed a significant change during MI training, we
tests comparing week 6 to week 10 were < 0.01 and MI/PP time 299
240 report only the result of the post-hoc comparison between the
ratios decreased by 28% and 33% (VMI and KMI), indicating an 300
241 first week (week 6) and last week (week 10) of training and the
evolution toward temporal equivalence for multiple-joint move- 301
242 percentage of change between these training weeks. When data
ments. 302
243 showed a significant change before and after MI training, we
Poor imagers exhibited changes in vividness for both VMI and 303
244 report 2 post-hoc comparisons — between the 3 baseline
KMI for single-joint movements (x2(4) = 26.2, P < 0.001, 304
245 measures (weeks 1, 3, 5) and each of the post-training measures
x2(4) = 19.7, P < 0.001) and multi-joint movements (x2(4) = 16.1, 305
246 (weeks 11 and 19).
P = 0.003, x2(4) = 39.4, P < 0.001). All 4 post-hoc tests comparing 306
247 Finally, to investigate whether training resulted in poor imagers
week 6 to week 10 had P-values < 0.01, indicating that vividness 307
248 reaching the same level of performance as good imagers, for each of
increased significantly for both imagery modalities and movement 308
249 the 5 training weeks, we compared MI vividness and MI/PP ratios
types. For single-joint movements, vividness scores increased by 309
250 between good and poor imagers by using the non-parametric Mann–
28% and 16% (VMI and KMI), whereas for multi-joint movements, 310
251 Whitney–Wilcoxon test. All statistical analyses were conducted by
scores increased by 31% and 41% (VMI and KMI). However, in 311
252 using R [30]. P < 0.05 was considered statistically significant.
contrast to good imagers, for poor imagers, vividness values at the 312
end of training were still below the range considered normal (blue 313
zone on the figures). Poor imagers displayed no changes in MI/PP 314
253 3. Results ratios for VMI or KMI during single-joint movements (x2(4) = 8.6, 315
P = 0.07, x2(4) = 3.9, P = 0.4) or multi-joint movements (x2(4) = 7.9, 316
254 3.1. MI training P = 0.1, x2(4) = 2.5, P = 0.7). 317
Comparison of MI ability across each week of MI training 318
255 All participants attended 15 training sessions (total session revealed that at each week, good imagers had significantly higher 319
256 duration 675 min). Participants with quadriplegia performed a vividness scores (all Ps < 0.05) and lower MI/PP ratios (all 320
257 total mean (SD) of 130 (22) actual movements and 1009 (337) Ps < 0.05) than poor imagers. Thus, even though both groups 321
258 imagined movements. The total mean (SD) duration was 23 (3) min improved with training, especially in terms of vividness, the poor 322
259 for actual movements and 206 (74) min for imagined movements. imagers never reached the same performance as good imagers. The 323
260 Control participants performed a total of 675 min of unsupervised week-by-week evolution for good and poor imagers is shown in 324
261 visualization. Table 2. 325

Please cite this article in press as: Mateo S, et al. Descriptive pilot study of vividness and temporal equivalence during motor imagery
training after quadriplegia. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.003
 G Model
REHAB 1192 1–9

S. Mateo et al. / Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx 5

Fig. 2. The time course of changes in vividness (top row) and temporal equivalence (bottom row) for a single joint movement (wrist extension) across 5 weeks of MI training in
individuals with quadriplegia. Box plots represent data from all participants. Visual MI is represented on the left and kinesthetic MI on the right. Vividness scores were
measured on a visual analog scale (VAS, 0–100 mm). Blue and red squares represent the mean (SEM) (for good and poor imagers) VAS score for all imagined movements
performed in that week. Temporal equivalence scores were measured as the average MI/PP ratio for movements performed in that week. Note that weekly averages are
displayed, but analyses were performed on data from the 15 sessions. P-values represent the results of the Linear Mixed Model (LMM) and asterisks represent the result of the
post-hoc tests comparing week 6 to week 10. *P  0.05; **P  0.01; ns: non-significant; MI: motor imagery; PP: physical practice; VMI: visual MI; KMI: kinesthetic MI.

326 3.3. Untrained movement: changes in temporal equivalence before P = 0.007), but poor imagers did not (x2(2) = 4.84, P = 0.08). Similar 340
327 and after training to the pattern observed in the whole group, for good imagers, ratios 341
decreased by 46% between the pre-training tests and week 11 but 342
328 Control participants had pre-training MI/PP ratios close to returned to initial levels at week 19 (post-hoc tests P = 0.01 and 343
329 temporal equivalence for the pointing task, and training visualiz- P = 1.0 respectively). 344
330 ing geometric shapes did not alter these ratios (x2(2) = 1.63,
331 P = 0.44), which were within the range of normal values at all
332 5 tests (Fig. 4). Participants with quadriplegia also had MI/PP pre- 4. Discussion 345
333 training ratios close to temporal equivalence, but MI training
334 significantly changed these ratios (x2(2) = 11.10, P = 0.004). Ratios Here we provide the first evidence that an MI training program 346
335 decreased by 38% between the pre-training tests and week 11 but in individuals with C6–C7 SCI improves functional equivalence. 347
336 returned to initial levels at week 19 (post-hoc tests P = 0.004 and Imagery vividness for the trained movements improved regardless 348
337 P = 0.6 respectively). The significant decrease between pre-training of the movement type or imagery modality. Because we previously 349
338 and week 11 was mostly due to the behavior of good imagers, who demonstrated that this training regimen improved tenodesis 350
339 exhibited significant changes in temporal equivalence (x2(2) = 9.9, kinematics in these same individuals, we suggest that improved 351

Please cite this article in press as: Mateo S, et al. Descriptive pilot study of vividness and temporal equivalence during motor imagery
training after quadriplegia. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.003
G Model
REHAB 1192 1–9

6 S. Mateo et al. / Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx

Fig. 3. The time course of changes in vividness (top row) and temporal equivalence (bottom row) for multi-joint movements (reach-to-grasp) across 5 weeks of MI training in
individuals with quadriplegia. Box plots represent data for all participants. Visual MI is represented on the left and kinesthetic MI on the right. Vividness scores were
measured by using a VAS (0–100 mm). Blue and red squares in the figure represent the mean (SEM) (for good and poor imagers) VAS score for all imagined movements
performed in that week. Temporal equivalence scores were measured as the mean MI/PP ratio for movements performed in that week. Note that weekly averages are
displayed, but analyses were performed on data for the 15 sessions. P-values represent the results of the LMM and asterisks represent the result of the post-hoc tests
comparing week 6 to week 10. *P  0.05; **P  0.01; ns: non-significant; MI: motor imagery; PP: physical practice; VMI: visual MI; KMI: kinesthetic MI.

Table 2
Vividness and MI/PP ratios for single- and multi-joint movements for good and poor imagers across the 5 training weeks.

VMI vividness KMI vividness VMI/PP ratio KMI/PP ratio

Movement Week Good Poor Good Poor Good Poor Good Poor

Single-joint 6 73 (41–100) 47 (19–76) 67 (44–89) 48 (5–91) 1.67 (0.04–3.29) 1.30 (0.77–1.83) 1.56 (0.02–3.14) 1.31 (0.65–1.98)
7 76 (50–100) 44 (27–61) 69 (48–91) 54 (27–82) 1.26 (0.68–1.84) 1.32 (0.71–1.94) 1.43 (0.65–2.22) 1.48 (0.62–2.34)
8 76 (53–99) 40 (29–51) 81 (66–97) 53 (48–58) 1.34 (0.41–2.28) 1.51 (0.95–2.07) 1.35 (0.63–2.06) 1.58 (0.88–2.29)
9 85 (65–100) 50 (38–61) 88 (75–100) 56 (42–70) 1.21 (1.07–1.37) 1.69 (0.54–2.83) 1.25 (0.96–1.56) 1.66 (0.60–2.71)
10 85 (67–100) 60 (46–73) 88 (76–99) 63 (51–74) 1.14 (0.52–1.78) 1.42 (0.59–2.25) 1.12 (0.45–1.78) 1.51 (0.49–2.52)
Multi-joint 6 72 (51–93) 55 (24–85) 66 (38–94) 46 (12–81) 1.43 (0.32–2.54) 1.37 (0.43–2.31) 1.38 (0.50–2.25) 1.44 (0.57–2.32)
7 71 (54–87) 61 (32–91) 71 (48–93) 62 (21–92) 1.07 (0.61–1.54) 1.23 (0.79–1.68) 1.32 (0.69–1.95) 1.39 (0.60–2.18)
8 81 (66–96) 57 (29–84) 82 (62–100) 61 (33–89) 1.14 (0.50–1.79) 1.54 (0.67–2.42) 1.17 (0.61–1.72) 1.57 (0.46–2.68)
9 86 (69–100) 61 (35–86) 86 (70–100) 61 (32–90) 0.88 (0.64–1.12) 1.39 (0.64–2.15) 1.02 (0.51–1.53) 1.51 (0.41–2.61)
10 83 (65–100) 64 (43–85) 83 (63–100) 65 (36–93) 0.87 (0.39–1.35) 1.32 (0.68–1.96) 0.93 (0.61–1.27) 1.39 (0.66–2.13)

Values shown are the mean of the 3 sessions in that week (95% confidence interval). Mann–Whitney–Wilcoxon tests comparing good and poor imagers for each modality,
movement type, and training week revealed that good imagers were consistently better than poor imagers (all Ps = 0.03). VAS: visual analog scale; MI: motor imagery; PP:
physical practice; VMI: visual MI; KMI: kinesthetic MI.

Please cite this article in press as: Mateo S, et al. Descriptive pilot study of vividness and temporal equivalence during motor imagery
training after quadriplegia. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.003
 G Model
REHAB 1192 1–9

S. Mateo et al. / Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx 7

Fig. 4. Functional equivalence on a complex pointing task measured before and after training in individuals with quadriplegia and in control participants. Box plots represent
data from all participants. Temporal equivalence on a non-trained pointing task before and after training in participants with C6–C7 quadriplegia (left panel) and aged-
matched healthy control participants (right panel). Good imagers are represented in blue squares, poor imagers in red, and control participants in black (mean [SEM]). The
week numbers indicate when the pointing task was tested. P-values represent the results of the LMM and asterisks represent the result of the post-hoc tests comparing the
pre-training tests to post-training tests at week 11 and week 19. *P  0.05; **P  0.01; NS: non-significant.

352 grasping after MI training is related to training-induced [34]. The absence of a post-training measure on the KVIQ means 380
353 improvements in MI quality. that we cannot know whether our training protocol improved 381
354 To our knowledge, no other studies have investigated whether general MI ability in individuals with quadriplegia. Thus, this 382
355 training alters the quality of imagined movements in individuals measure should be incorporated into future studies. However, our 383
356 with quadriplegia. One study-examining individuals with stroke study is the first to show that an MI training that led to improved 384
357 and neglect showed that kinesthetic MI training increased KMI tenodesis kinematics increased the vividness of those movements 385
358 vividness scores on the revised MI Questionnaire (MIQ) completed imagined during the training. 386
359 before and after training but did not assess the quality of the Regardless of initial imagery ability, all individuals with 387
360 movements imagined throughout training [31]. Similarly, a study quadriplegia benefited from training, although poor imagers never 388
361 of golf shot performance in healthy individuals reported that MI caught up to good imagers. Good imagers had vividness scores 389
362 training improved the ease of generating both VMI and KMI (using within the normal range by the end of training and were the only 390
363 the revised MIQ), but the authors did not provide any information group showing improved MI/PP ratios (during multi-joint move- 391
364 about changes in the quality of the movements that were imagined ments). Good imagers may be close to a ceiling in their vividness 392
365 during the training [32]. In another case–control study, 30 min of performance, which enabled them to devote cognitive resources to 393
366 visual and kinesthetic imagery of the Nine Hole Peg Test resulted in achieving temporal equivalence. Indeed, their MI/PP ratios 394
367 faster PP times for this test as well as for a non-trained, mirror improved from 1.4 to 0.9 (for multi-joint movements), which is 395
368 version of the Nine Hole Peg Test. Furthermore, self-rated vividness close to the functional equivalence ratio of 1.1 reported in healthy 396
369 scores for imagined movements in the 2 final training blocks were individuals [35] or individuals with lower-limb amputation [33] 397
370 positively correlated with the increase in speed of physically and falls within the range considered normal (0.75–1.25) for 398
371 practicing the trained task [19]. A study of lower-leg amputees individuals after a stroke [29]. This idea is consistent with evidence 399
372 turned the question around and asked not whether MI training showing that experts show focused and efficient organization of 400
373 alters imagery ability and/or PP but whether improvements in PP task-related neural networks, whereas novices have difficulty 401
374 (arising from participation in a rehabilitation program) alter MI focusing and filtering out irrelevant information [36]. Indeed, poor 402
375 ability. Using the average KVIQ score for 4 lower-limb movements, imagers might have focused exclusively on generating the image or 403
376 the authors reported improvement in MI ability after successful kinesthetic sensation, whereas the more ‘‘expert’’ good imagers 404
377 rehabilitation [33]. Together with the present results, these studies might have been able to switch their focus to the goal of making the 405
378 advocate that imagery ability and physical performance can be movement [37]. These results suggest that clinicians should 406
379 improved by MI training and that rehabilitation can alter MI ability monitor MI vividness and that when a certain level of expertise is 407

Please cite this article in press as: Mateo S, et al. Descriptive pilot study of vividness and temporal equivalence during motor imagery
training after quadriplegia. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.003
 G Model
REHAB 1192 1–9

8 S. Mateo et al. / Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx

408 achieved, attention to movement duration and temporal equiva- of vividness. From these preliminary results, we propose several 471
409 lence could further increase the benefits of MI training. recommendations for clinical practice and future research. In a 472
410 In addition to assessing changes in MI quality during training, clinical setting, if only one parameter of MI training can be 473
411 we also investigated the transfer/generalization to an untrained monitored, it should be vividness. Furthermore, it could be useful 474
412 task by assessing temporal equivalence during a non-trained to measure imagery ability before training and to tailor the training 475
413 pointing task before and after training. We found some evidence accordingly (e.g., asking poor imagers to focus on vividness and 476
414 for transfer in that MI/PP ratios decreased immediately after good imagers to focus on both vividness and temporal equiva- 477
415 training, particularly for good imagers. This finding could be lence). Finally, although it remains to be tested, good imagers 478
416 interpreted as further evidence that good imagers were able to might benefit from receiving feedback about the duration of their 479
417 control multiple parameters, but their final MI/PP ratios were imagined movements, which could increase the effects of MI 480
418 below the normal range, because they underestimated the MI training on temporal equivalence and in turn lead to further 481
419 duration as compared with PP duration [29,33,35]. The pointing improvements in physical practice. 482
420 task may have been easier than the trained grasping tasks and led
421 good imagers to underestimate its duration when they imagined it. Author contributions 483
422 We did not emphasize movement speed during our training but
423 instead told participants to move at a comfortable pace. Hence, not SM, GR and CC provided idea/research design. All the authors 484
424 unsurprisingly, we observed no change in the speed of physically were involved in writing. SM provided data collection and analysis. 485
425 practiced movements [7,38]; however, studies that instructed All authors provided project management and consultation 486
426 participants to imagine moving as fast as possible reported (including review of the manuscript before submission). GR 487
427 decreased duration of physically practiced movements [19]. All provided participants. GR and CC provided facilities/equipment 488
428 6 participants with quadriplegia had stable tenodesis grasp and institutional links. 489
429 kinematics before MI training, showing that they had learned
430 the tenodesis grasp during their rehabilitation. Kinematics and Disclosure of interest 490
431 brain imaging changes reported in response to MI training are
432 likely related to a second phase of motor learning [7]. Future The authors declare that they have no competing interest. 491
433 studies designed specifically for that purpose could more clearly
434 elucidate the effect of motor learning during MI training (e.g., [39]).
Uncited reference Q3 492
435 Because we conducted a retrospective analysis of data from a
436 study designed to examine tenodesis kinematics and brain
[24]. 493
437 plasticity, several aspects of the current report could be improved
438 when conducting future studies of MI in individuals with
Acknowledgements 494
439 quadriplegia. First, we did not have a questionnaire-based
440 assessment of MI ability (e.g., KVIQ) before and after training or
441 a measure of changes in imagery vividness during the non-trained We thank F. Di Rienzo (PhD) and Professor A. Guillot for their 495
442 pointing task, which prevents a complete discussion about the help in the study design. This research was supported by a grant 496
443 generalization of MI vividness improvement. Similarly, the from the Hospital Clinical Research Program (HCRP-2008-541/42), 497
444 inclusion of other mental chronometry assessments before and from the Nursing and Paramedical Hospital Research Program 498
445 after training (e.g., the imagined Box and Block Test [40]), would (PHRIP-16-0605), from the Labex/Idex (ANR-11-LABX-0042) and 499
446 have made it easier to tease apart the specific effects of training IHU CeSaMe (ANR-10-IBHU-0003). Part of this work has been 500
447 from general improvement in MI ability. Furthermore, most of the presented at the international congress of the World Congress of 501
448 effects we report were similar for both visual and kinesthetic Neurorehabilitation in Philadelphia (USA), May 11–13, 2016. 502
449 imagery, but future studies should randomize the order of visual
450 and kinesthetic imagery to avoid confounding the effects of fatigue
451 and training order. Finally, since our study was the first to examine Appendix A. Supplementary data 503
452 the possible beneficial effect of MI on tenodesis grasp, we
453 maximized our chances of finding a positive outcome by including Supplementary data associated with this article can be found, in 504
the online version, at http://www.dx.doi.org.10.1016/j.plantsci. 505
454 both wrist extension and reach-to-grasp movements in the
455 training regimen. The downside of this choice is that we cannot 2004.08.011. 506
456 separate out the effect of the task practiced (wrist extension with
457 no object to be grasped or reaching to grasp with a palmar grip or References 507
458 with a lateral grip) and the training week. In the future, it will be
[1] Kirshblum SC, Burns SP, Biering-Sorensen F, Donovan W, Graves DE, Jha A, et al. 508
459 useful to know whether both movement types were necessary for International standards for neurological classification of spinal cord injury 509
460 the tenodesis grasp improvements or whether simplifying training (Revised 2011). J Spinal Cord Med 2011;34:535–46. 510
461 (e.g., to include wrist extension only) could be an option. [2] Long CII, Lawton EB. Functional significance of spinal cord lesion level. Arch 511
Phys Med Rehabil 1955;36:249–55. 512
[3] Kirshblum SC, Priebe MM, Ho CH, Scelza WM, Chiodo AE, Wuermser L-A. Spinal 513
Cord Injury Medicine. 3. Rehabilitation phase after acute spinal cord injury. 514
462 5. Conclusion Arch Phys Med Rehabil 2007;88:S62–70. 515
[4] Woolsey RM. Rehabilitation outcome following spinal cord injury. Arch Neurol 516
1985;42:116–9. 517
463 The results of this descriptive pilot study show that a 5-week MI [5] Mateo S, Revol P, Fourtassi M, Rossetti Y, Collet C, Rode G. Kinematic charac- 518
464 training program that improved tenodesis grasp kinematics [7] teristics of tenodesis grasp in C6 quadriplegia. Spinal Cord 2013;51:144–9. 519
465 also affected functional equivalence. Individuals who were good [6] Mateo S, Roby-Brami A, Reilly K, Rossetti Y, Collet C, Rode G. Upper limb 520
kinematics after cervical spinal cord injury: a review. J Neuroeng Rehabil 521
466 imagers at the onset of training showed improvement in both the 2015;12:9. 522
467 vividness and functional equivalence of trained movements, [7] Mateo S, Di Rienzo F, Reilly KT, Revol P, Delpuech C, Daligault S, et al. 523
468 whereas poor imagers showed only improved vividness. These Improvement of grasping after motor imagery in C6–C7 tetraplegia: a kine- 524
matic and MEG pilot study. Restor Neurol Neurosci 2015;33:543–55. 525
469 results suggest that individuals primarily control vividness but can [8] Mateo S, Di Rienzo F, Bergeron V, Guillot A, Collet C, Rode G. Motor imagery 526
470 also target temporal equivalence once they achieve a certain level reinforces brain compensation of reach-to-grasp movement after cervical 527

Please cite this article in press as: Mateo S, et al. Descriptive pilot study of vividness and temporal equivalence during motor imagery
training after quadriplegia. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.003
 G Model
REHAB 1192 1–9

S. Mateo et al. / Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx 9

528 spinal cord injury. Front Behav Neurosci 2015;9:1–12. http://dx.doi.org/ [24] Decety J, Boisson D. Effect of brain and spinal cord injuries on motor imagery. 569
529 10.3389/fnbeh.2015.00234. Eur Arch Psychiatry Clin Neurosci 1990;240:39–43. 570
530 [9] Decety J, Grèzes J. Neural mechanisms subserving the perception of human [25] Feltz DL, Landers DM. The effects of mental practice on motor skill learning and 571
531 actions. Trends Cogn Sci 1999;3:172–8. performance: a meta-analysis. J Sport Psychol 1983;5:25–57. 572
532 [10] Di Rienzo F, Guillot A, Daligault S, Delpuech C, Rode G, Collet C. Motor [26] Ditunno JJ, Cohen ME, Hauck WW, Jackson AB, Sipski ML. Recovery of upper- 573
533 inhibition during motor imagery: a MEG study with a quadriplegic patient. extremity strength in complete and incomplete tetraplegia: a multicenter 574
534 Neurocase 2013;20:524–39. study. Arch Phys Med Rehabil 2000;81:389–93. 575
535 [11] Hanakawa T, Dimyan MA, Hallett M. Motor planning, imagery, and execution [27] Oldfield R. The assessment and analysis of handedness: the Edinburgh inven- 576
536 in the distributed motor network: a time-course study with functional MRI. tory. Neuropsychologia 1971;9:97–113. 577
537 Cereb Cortex 2008;18:2775–88. [28] Bohannon R, Smith M. Interrater reliability of a modified Ashworth scale of 578
538 [12] Grush R. The emulation theory of representation: motor control, imagery, and muscle spasticity. Phys Ther 1987;67:206–7. 579
539 perception. Behav Brain Sci 2004;27:377–96 [Discussion 396–442]. [29] McInnes K, Friesen C, Boe S. Specific brain lesions impair explicit motor 580
540 [13] Guillot A, Collet C. Duration of mentally simulated movement: a review. J imagery ability: a systematic review of the evidence. Arch Phys Med Rehabil 581
541 Motor Behav 2005;37:10–20. 2016;97:478–89 [e1]. 582
542 [14] Munzert J, Lorey B, Zentgraf K. Cognitive motor processes: the role of motor [30] R Core Team. R: a language and environment for statistical computing. Vienna, 583
543 imagery in the study of motor representations. Brain Res Rev 2009;60:306–26. Austria: R Foundation for Statistical Computing; 2013. 584
544 [15] Driskell JE, Copper C, Moran A. Does mental practice enhance performance? J [31] Leifert-Fiebach G, Welfringer A, Babinsky R, Brandt T. Motor imagery training 585
545 Appl Psychol 1994;79:481–92. in patients with chronic neglect: a pilot study. NeuroRehabilitation 586
546 [16] Lequerica A, Rapport L, Axelrod BN, Telmet K, Whitman DR. Subjective and 2013;32:43–58. 587
547 objective assessment methods of mental imagery control: construct valida- [32] Wright DJ, McCormick SA, Birks S, Loporto M, Holmes PS. Action observation 588
548 tions of self-report measures. J Clin Exp Neuropsyc 2002;24:1103–16. and imagery training improve the ease with which athletes can generate 589
549 [17] Malouin F, Richards CL, Jackson PL, Lafleur MF, Durand A, Doyon J. The imagery. J Appl Sport Psychol 2015;27:156–70. 590
550 Kinesthetic and Visual Imagery Questionnaire (KVIQ) for assessing motor [33] Saruco E, Guillot A, Saimpont A, DI Rienzo F, Durand A, Mercier C, et al. Motor 591
551 imagery in persons with physical disabilities: a reliability and construct imagery ability of patients with lower-limb amputation: exploring the course 592
552 validity study. J Neurol Phys Ther 2007;31:20–9. of rehabilitation effects. Eur J Phys Rehabil Med 2017[Epub ahead of print]. Q4593
553 [18] Roure R, Collet C, Deschaumes-Molinaro C, Delhomme G, Dittmar A, Vernet- [34] Cumming J, Williams SE. The role of imagery in performance. Handbook of 594
554 Maury E. Imagery quality estimated by autonomic response is correlated to Sport and Performance Psychology; 2012;213–32. 595
555 sporting performance enhancement. Physiol Behav 1999;66:63–72. [35] Malouin F, Richards CL, Durand A. Slowing of motor imagery after a right 596
556 [19] Ruffino C, Papaxanthis C, Lebon F. The influence of imagery capacity in motor hemispheric stroke. Stroke Res Treat 2012;2012:297217. 597
557 performance improvement. Experimental Brain Research 2017;235:3049–57. [36] Milton J, Solodkin A, Hluštı́k P, Small SL. The mind of expert motor perfor- 598
558 [20] Dickstein R, Deutsch JE. Motor imagery in physical therapist practice. Phys mance is cool and focused. NeuroImage 2007;35:804–13. 599
559 Ther 2007;87:942–53. [37] Milton J, Small SL, Solodkin A. Imaging motor imagery: methodological issues 600
560 [21] Jackson PL, Lafleur MF, Malouin F, Richards C, Doyon J. Potential role of mental related to expertise. Methods 2008;45:336–41. 601
561 practice using motor imagery in neurologic rehabilitation. Arch Phys Med [38] Di Rienzo F, Guillot A, Mateo S, Daligault S, Delpuech C, Rode G, et al. 602
562 Rehabil 2001;82:1133–41. Neuroplasticity of prehensile neural networks after quadriplegia. Neurosci- 603
563 [22] Di Rienzo F, Collet C, Hoyek N, Guillot A. Impact of neurological deficits on ence 2014;22:82–92. 604
564 motor imagery: a systematic review of clinical evaluations. Neuropsychol Rev [39] Sobierajewicz J, Przekoracka-Krawczyk A, Jaśkowski W, Verwey WB, van der 605
565 2014;24:116–47. Lubbe R. The influence of motor imagery on the learning of a fine hand motor 606
566 [23] Malouin F, Richards CL, Durand A, Descent M, Poire D, Fremont P, et al. Effects skill. Exp Brain Res 2017;235:305–20. 607
567 of practice, visual loss, limb amputation, and disuse on motor imagery vivid- [40] Greiner J, Schoenfeld MA, Liepert J. Assessment of mental chronometry (MC) in 608
568 ness. Neurorehabil Neural Repair 2009;23:449–63. healthy subjects. Arch Gerontol Geriat 2014;58:226–30. 609

Please cite this article in press as: Mateo S, et al. Descriptive pilot study of vividness and temporal equivalence during motor imagery
training after quadriplegia. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.003