Neuroscience 273 (2014) 189–198


Key words: force steadiness, spectral analysis, sample entropy, knee extensors, hemiparesis.

Abstract—We examined changes in the variability, frequency composition, and complexity of force signal from subacute to chronic stage of stroke during maintenance of isometric knee extension and compared these parameters between chronic stroke and healthy subjects. The sample included 15 healthy (65 ± 8 years) and 23 chronic stroke subjects (65 ± 14 years, 6–112 months post-stroke) of whom 10 (64 ± 15 years) were also examined 11–22 days post-stroke (subacute stage). The subjects performed isometric knee extension at 10%, 20%, 30%, and 50% of peak torque for 10 s (two trials each). Coefficient of variation (CV) was used as a measure of force variability. The median frequency and relative power in the 0–3, 4–6, and 8–12 Hz bands were obtained through a power spectrum analysis of the force signal. The signal complexity was quantified using the sample entropy (SampEn). The longitudinal analysis revealed a significant decrease in CV from subacute to chronic stage across all contraction levels (P < 0.001) but no significant changes in the frequency and entropy parameters. Comparison between the chronic stroke and control subjects revealed no significant difference in CV across the force levels (P > 0.05) but significantly decreased median frequency (P < 0.01), with the relative power increased in 0–3 Hz band and decreased in 4–6 and 8–12 Hz bands in both paretic and non-paretic legs (P < 0.001). SampEn was also significantly decreased in chronic stroke, bilaterally (P < 0.001). These results indicate a shift toward lower frequencies and a less complex physiological process underlying force control in chronic stroke. The overall results suggest the improvement in force variability from subacute to chronic stroke but without normalization in the frequency composition and complexity of the force signal. Thus, disordered structure of the force signal remains a marker of impaired motor control long after stroke occurrence despite apparent recovery in force variability. Ó 2014 IBRO. Published by Elsevier Ltd. All rights reserved.

INTRODUCTION The ability to produce a steady force is impaired in stroke patients. Lodha et al. (2010) reported a significantly greater force variability (coefficient of variation (CV)) in the paretic wrist/finger extensors for nine stroke subjects (4 months to 12 years post-onset) compared to nine controls at 5%, 25%, and 50% of the maximum voluntary contraction (MVC). The greatest difference between stroke and control subjects was found for bilateral tasks at 5% and 50% of MVC (Lodha et al., 2012). Chow and Stokic (2011) examined 33 subjects within a month of stroke on the isometric knee extension task at 10%, 20%, 30%, and 50% of MVC and reported significantly increased CV in both paretic and non-paretic legs of stroke subjects compared to controls across all force levels. In addition to simple measures of force variability (CV), nonlinear analytic approaches allow examination of the structure of force signal and provide an insight into underlying physiological processes (Schiffman et al., 2006). The structure of a time series includes both time (complexity analysis) and frequency (power spectrum analysis) domains. Approximate entropy, a measure of complexity structure of the force signal, was significantly decreased in stroke subjects during a constant wrist/finger extension task, particularly at higher force levels (Lodha et al., 2010). These investigators ascribed the less complex force signal to the lack of motor adaptability associated with relatively fixed or stereotypic patterns in motor coordination and abnormal movement synergies in chronic stroke. Chow and Stokic (2013) performed a power spectrum analysis on force signals during a constant knee extension task in subacute stroke and reported a shift from 4–12 Hz to 0–3 Hz. A shift toward lower frequencies within the 0–1 Hz band has also been reported in chronic stroke in an isometric grip task (Lodha et al., 2013). The predominance of lower frequencies in the power spectrum and decreased entropy during constant-force tasks have been found in Down syndrome (Heffernan et al., 2009) and Parkinson’s disease (Vaillancourt et al., 2001), suggesting disordered structure of the force signal across

*Corresponding author. Address: Methodist Rehabilitation Center, 1350 East Woodrow Wilson Drive, Jackson, MS 39216, USA. Tel: +1-601-364-3402; fax: +1-601-364-3305. E-mail address: [email protected] (J. W. Chow). Abbreviations: ANOVA, analysis of variance; CV, coefficient of variation; FFT, Fast Fourier Transform; FM, Fugl-Meyer scale; MVC, maximum voluntary contraction; RMI, Rivermead Mobility Index; SampEn, Sample entropy; SD, standard deviation. 0306-4522/Ó 2014 IBRO. Published by Elsevier Ltd. All rights reserved. 189


J. W. Chow, D. S. Stokic / Neuroscience 273 (2014) 189–198

different neurological disorders. In healthy subjects, 0–3 Hz band during a constant-force task has been associated with visuomotor processing (Freund and Hefter, 1993; Slifkin et al., 2000; Vaillancourt et al., 2001), 4–6 Hz band with long-latency stretch reflexes (Marsden, 1978; McAuley and Marsden, 2000), and 8–12 Hz band with short-latency stretch reflexes (Marsden, 1978; McAuley and Marsden, 2000). Our previous findings (Chow and Stokic, 2013) raise a question whether the increased force variability and altered force frequency characteristics observed in the subacute stage of stroke carry over into the chronic stage. Also it is unknown whether decreased complexity of the force signal in chronic stroke reported for the wrist/finger extensors (Lodha et al., 2010) also applies to the knee extensors. Therefore, the first aim of this study was to examine changes in the variability, frequency composition, and complexity of the force signal from the subacute (within the first month of stroke) to the chronic stage of stroke (at least 6 months post-stroke) during an isometric knee extension task. We hypothesized some degree of normalization of CV, frequency composition, and complexity of the force signal over time (hypothesis 1). Because the force structure of an isometric knee extension has not been previously investigated in chronic stroke, our second aim was to compare variability, frequency composition, and complexity of the force signal between persons with chronic stroke and healthy controls. Based on the previous findings (Lodha et al., 2010, 2013; Chow and Stokic, 2013), we hypothesized that chronic stroke subjects would show an increase in force variability, a shift toward lower frequencies in the power spectrum, and a decrease in complexity of the force signal in both the paretic and non-paretic legs compared to controls (hypothesis 2). Since the associations between force parameters and clinical measures of motor recovery were not directly related to the tested hypotheses, these correlations were explored in secondary analyses.

EXPERIMENTAL PROCEDURES Subjects Twenty-three community-dwelling persons with chronic stroke were included in this study (Table 1). The inclusion criteria were at least 6 months post-stroke, single unilateral stroke or multiple strokes on the same side, unimpaired vision (able to see a line on a monitor), able to extend both knees against gravity in the seated position, and able to follow simple instructions. Those with clinical evidence of visual (hemianopia) or perceptual (neglect) deficits, heart diseases, uncontrolled hypertension, normal pressure hydrocephalus, knee pain, or artificial knee replacement were excluded. The control group included 15 subjects (age 65 ± 8 years, height 177 ± 11 cm, body mass 82 ± 14 kg, 11 men) with normal or corrected-to-normal vision and no reported orthopedic or neurological disorders at the time of testing. The age difference between the two groups was not significant (unpaired t-test, P = 0.63). All subjects signed the informed consent approved by the institutional review board. Prior to force tasks, stroke subjects were

assessed by the same physical therapist on the lower extremity motor section of the Fugl-Meyer (FM) scale (range 0–34, Fugl-Meyer et al., 1975), modified Ashworth scale (range 1–5, Bohannon and Smith, 1987), and Rivermead Mobility Index (RMI) (maximum 15, Collen et al., 1991; Hsieh et al., 2000) (not collected in two subjects because the therapist was not available). Ten stroke subjects were tested twice – within the first month of stroke just prior to discharge from in-patient rehabilitation and then at 6–8 months post-stroke (Table 1). They received outpatient physical and occupational therapy for up to 3 months after the inpatient discharge.

Experimental setup and protocol With the subject in a seated position, knee extension torques were collected using a Biodex System 3 isokinetic dynamometer (Biodex Medical Systems, Inc., New York, NY, USA) and a custom-built amplifier connected directly to the torque sensor of the dynamometer (overall sensitivity 57.5 mV/Nm). Torque signals from the dynamometer were fed to a 17-in. LCD monitor that was mounted on a swing arm and an EvaRT data acquisition system (Motion Analysis Corp., Santa Rosa, CA, USA, sample rate 1200 Hz, 12-bit analog-to-digital resolution). Before practice trials, the monitor was positioned according to individual preferences, usually 40–50 cm directly in front of the head. Both legs of stroke patients were tested in a random order and only the self-reported dominant (preferred ball kicking) leg in controls. The warm-up included five repetitions of maximum isokinetic knee extension-flexion at 210°/s and 60°/s. The subject then performed three to four trials of maximum isometric knee extension (90° angle, 3–4 s each, 1-min pause). The largest force (proportional to the largest torque because of the constant moment arm) was used as the MVC. After several practice trials, the subject was asked to extend the knee, match the displayed torque signal (a horizontal line) with a designated target force marked on the monitor (10%, 20%, 30%, or 50% of MVC), and maintain the force for 10 s, as accurately and steadily as possible. The four force levels were presented in a random order and two trials per level were completed with a 30–60-s rest in between.

Data analyses MVC torques were smoothed using a sliding average of 600 data points (0.5-s window). Torques during constant-force trials were filtered using a second-order Butterworth low-pass filter with 30-Hz cutoff. The CV, calculated as the ratio between standard deviation (SD) and mean torque (CV = SD/mean  100%), was used to quantify force variability (steadiness). Only the middle 8 s of each 10-s trial were analyzed to exclude the ramp up and down portions of the force signal. Out of two trials collected at each force level, the one with a lower CV was used for statistical analysis. The ratio of paretic


J. W. Chow, D. S. Stokic / Neuroscience 273 (2014) 189–198 Table 1. Subject characteristics

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Mean SD


Age (years)

Height (cm)

Mass (kg)

Chronic postinjury (months)

Subacute postinjury (days)

Paretic side

Lower extremity Fugl-Meyer*

Rivermead Mobility Index*

MVC Asymmetry*


64 88 79 53 77 57 56 61 58 75 80 43 53 75 82 73 62 87 55 56 39 50 62

163 157 178 165 155 163 180 175 183 170 191 173 165 173 175 183 157 173 165 179 160 168 152

55 55 80 59 64 68 75 80 98 75 94 90 78 81 71 120 86 75 88 83 81 91 55

112 16 12 14 12 14 51 15 13 18 12 14 6 7 6 7 6 6 7 6 8 6 8

n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a 16 12 11 14 14 15 17 14 12 22


n/a 28 34 34 24 17 18 12 29 21 25 27 29 33 (n/a) 34 (30) 32 (31) 24 (19) 32 (30) 17 (22) 19 (16) 29 (28) 20 (14) 34 (30)

n/a 15 15 15 15 13 7 14 14 11 14 15 13 13 (n/a) 15 (13) 13 (14) 9 (5) 14 (13) 15 (13) 13 (13) 15 (15) 12 (11) 15 (15)

0.60 0.57 0.69 1.06 0.98 0.28 0.38 0.74 0.68 0.71 0.80 0.97 0.75 0.79 0.79 0.85 0.57 1.01 0.52 0.79 0.76 0.43 0.88

64.9 13.6

169.8 10.0

26.0 (27.4) 6.7 (6.7)

13.4 (13.4) 2.1 (1.9)

0.73 (0.76) 0.21 (0.19)

78.7 15.7

16.6 22.7

14.7 3.2

(0.55) (0.80) (0.78) (0.66) (0.85) (0.55) (0.14) (0.92) (0.55) (0.62)

Abbreviations: M, male; F, female; R, right; L, left; MVC, maximum voluntary contraction; n/a, not available/not applicable. * Values for subacute stage in parentheses, if applicable.

to non-paretic MVC torque was used as an index of strength asymmetry (MVC asymmetry, Table 1). The Fast Fourier Transform (FFT) function in Matlab (MathWorks, Inc., Natick, MA, USA) was used to obtain the power spectrum of the torque signals. The time constant was 8 s and the frequency bin resolution was 0.125 Hz (1200 Hz/9600 samples). We focused on the power between 0 and 12 Hz because the remaining power in the 12–20-Hz and 20–30-Hz bands accounted for only 1.08 ± 2.29% and 0.22 ± 0.51% of the total power, respectively, consistently across all subjects, limbs and force levels. Parameters extracted from the power spectrum of each trial were the median frequency (the frequency that divides the 0–12-Hz spectrum into two equal areas) and relative proportion of power in the 0–3 Hz (low-frequency band), 4–6 Hz (mid-frequency band) and 8–12 Hz (high-frequency band). The latter were computed as the ratio (%) between the integral of the power spectrum in each respective band and the integral of the power spectrum in the 0–12-Hz band (Kouzaki et al., 2004; Dewhurst et al., 2007; Muceli et al., 2011). Unlike the FFT analysis, the entropy analysis is sensitive to the sampling rate. To ensure our findings are comparable to the literature, the torque signals were down-sampled to 100 Hz before entropy computation. For assessing the complexity of knee extension force in the time domain, we chose sample entropy (SampEn) (Richman and Moorman, 2000) over the more commonly used approximate entropy (Pincus, 1991) because

SampEn has greater relative consistency and is less dependent on the dataset length (Richman and Moorman, 2000). SampEn (m, r, N) is the negative natural logarithm of the conditional probability that a dataset of length N, having repeated itself within a tolerance r for m points, will also repeat itself for m + 1 points, without allowing self matches (Lake et al., 2002). The analysis was completed using a Matlab program downloaded from the open source PhysioNet ( with N = 800, m = 3, and r = 0.2 SD of the time series (Lake et al., 2002; Simmons et al., 2012). SampEn values range from 0 (less complex) to 2 (random noise). Statistical analysis For hypothesis 1 (subacute to chronic comparison), Friedman’s non-parametric 2-way analysis of variance (ANOVA) by ranks (two Time points in columns and four Force Levels in rows) was performed on each force outcome (i.e., comparison of Time effects without testing for Force Level effects or interactions). Prior to testing hypothesis 2, we first determined if the outcomes significantly differed between the 10 subjects studied twice and 13 subjects studied in the chronic stage only because the time from stroke occurrence to evaluation was borderline significant between the two groups (7 ± 1 vs. 24 ± 29 months, unequal variance t-test, P = 0.051). Unpaired t-test and 2 Group  4 Force Level ANOVA with adjustment per Levene’s and Mauchly’s tests were used, as appropriate. No


J. W. Chow, D. S. Stokic / Neuroscience 273 (2014) 189–198

significant differences were found for FM motor score, RMI, MVC torque, and MVC asymmetry (t-tests P P 0.353). Also, there were no significant main effects of Group (P P 0.087) or Group  Force Level interactions (P P 0.289) for the CV, frequency, and entropy parameters in either leg. This justified the pooling of all data for testing hypothesis 2. For this, the paretic and non-paretic legs were separately compared to the control leg using a 2 (Leg: paretic/non-paretic, control)  4 (Force Level: 10%, 20%, 30%, 50% MVC) ANOVA with repeated measures on the second factor and Greenhouse–Geisser correction (Mauchly’s test of sphericity, P 6 0.05) for each force outcome. Because the emphasis was on between-leg comparisons and interactions, the significant main effect of Force Level was ignored to reduce the number of statistical tests (Chow and Stokic, 2013). In secondary analyses, associations between the MVC asymmetry, FM motor score, and RMI on the one side and the frequency and entropy parameters on the other side (paretic leg only) were explored at each contraction level using both linear and non-linear models. The relative spectral power in 0–3-Hz band was best fitted against FM score with an adjungated hyperbolic function [Y = a/(X b) + c, where a, b, and c are constants], whereas the relative spectral power in the other two frequency bands and FM score was best fitted with an exponential growth function [Y = d  exp(e  X), where d and e are constants]. Considering the number of statistical tests performed, the P-value was set at a more stringent level of 0.01.

RESULTS All stroke subjects were ambulatory (RMI P 7) at the chronic stage with FM motor scores from 12 to 34 on the paretic side (Table 1). Ashworth scores in the paretic knee extensors were 1 in all but one case (score of 2), indicating no clinical hypertonia. The MVC torques were significantly lower (P < 0.001, unpaired t-test) in the paretic leg (99 ± 45 Nm) compared to either the non-paretic (138 ± 51 Nm) or control leg (164 ± 57 Nm). The difference remained significant (P < 0.001) after MVC torque was normalized to the body mass (paretic leg 1.25 ± 0.45 Nm/kg, non-paretic leg 1.74 ± 0.47 Nm/kg, control 1.98 ± 0.58 Nm/kg). The 10 stroke subjects assessed longitudinally showed motor improvements from the subacute to chronic stage that approached significance (FM motor score 24.4 ± 6.7 to 27.4 ± 6.7, paired t-test, P = 0.063; RMI 12.4 ± 3.0 to 13.4 ± 1.9, P = 0.081; paretic MVC torque 93 ± 50 to 110 ± 40 Nm, P = 0.102; non-paretic MVC torque 142 ± 54 to 150 ± 52 Nm, P = 0.401; MVC asymmetry 0.64 ± 0.22 to 0.76 ± 0.19, P = 0.102). As to hypothesis 1, the only significant change from the subacute to chronic stage was a decrease in CV in the paretic leg (P < 0.001) with a similar trend in the non-paretic leg (P = 0.028) (Figs. 1 and 2). The average decrease in CV across all force levels in the paretic leg was from 5.4 ± 3.6% at subacute stage to 3.9 ± 3.7% at chronic stage (Fig. 2). The corresponding values for the non-paretic leg were 2.3 ± 1.0% and

Fig. 1. Representative torque and power spectrum profiles for the paretic and non-paretic legs of a stroke subject and the dominant leg of a control subject at different force levels illustrating the shift toward lower frequencies and decreased complexity in the force structure after stroke. Dashed lines indicate target forces. Note the same 20-Nm range in torque plots and variable ranges in power spectrum plots. Abbreviations: CV, coefficient of variation; SampEn, sample entropy; MF, median frequency (Hz), B1: 0–3 Hz band, B2: 4–6 Hz band, B3: 8–12 Hz band. The sum of B1, B2, and B3 is less than 100% because of the gaps between adjacent frequency bands.

J. W. Chow, D. S. Stokic / Neuroscience 273 (2014) 189–198


Fig. 2. Comparison of coefficients of variations and different frequency and entropy parameters across four force levels between the subacute stage (filled symbols) and chronic stage (unfilled symbols) of stroke for the paretic (black) and non-paretic (gray) legs (n = 10). The error bars indicate standard deviations. ⁄Significant difference between the paretic and non-paretic leg (P < 0.001).

1.8 ± 0.7%, respectively. Changes in the spectral frequency and entropy parameters were not significant (P P 0.174). In the chronic stage (hypothesis 2), all force variables (P 6 0.002) except CV (P P 0.119) were significantly different when paretic or non-paretic leg was compared with control (Fig. 3). Specifically, both legs of stroke subjects showed significantly decreased median frequency, increased relative power in 0–3-Hz band, decreased relative power in 3–6 and 8–12-Hz bands, and decreased SampEn. In the paretic leg, significant Leg x Force Level interactions were found for the relative power in 0–3 and 8–12-Hz bands as well as SampEn (P 6 0.006), whereas in the non-paretic leg the only significant Leg  Force Level interaction was for the relative power in 8–12-Hz band (P = 0.007). The interaction plots (Fig. 3) indicate that, with increasing force level, there was a smaller rate of decrease in the relative power in 0–3-Hz band and a smaller rate of increase in the relative power in 8–12-Hz band and SampEn in the paretic than control leg. Only the rate of increase in the relative power in 8–12-Hz band was smaller in the non-paretic than control leg.

In terms of correlations with the force parameters in the paretic leg, MVC asymmetry did not correlate with the CV, frequency, or entropy measures (P P 0.093). However, significant non-linear correlations were found between FM motor score and the relative power in 0–3 and 4–6-Hz bands at 20–50% force level (P 6 0.005, Fig. 4). A significant correlation was also found between FM score and the relative power in 8–12-Hz band at the 30% force level (P = 0.006). The significant correlation between RMI and the relative power in 8–12-Hz band at 10% MVC (r = -0.56, P = 0.006) was confirmed to be due to two outliers.

DISCUSSION In this study, we examined whether increased variability and altered structure of the isometric force signal found in knee extensors within a month of stroke (Chow and Stokic, 2013) are still present at 6 months or later and how that relates to motor recovery. For this, we longitudinally followed-up a subset of stroke subjects from the subacute to chronic stage (hypothesis 1) and compared in a cross-section study chronic stroke to healthy subjects


J. W. Chow, D. S. Stokic / Neuroscience 273 (2014) 189–198

Fig. 3. Comparison of coefficients of variations and different frequency and entropy parameters across four force levels between chronic stroke and healthy subjects. The paretic or non-paretic leg of stroke subjects was compared to controls using a 2 (Leg)  4 (Force Level, FL) ANOVA. P-values in parenthesis refer to significant main effects of Leg and significant Leg  FL interactions (P 6 0.01). The error bars indicate standard deviations.

(hypothesis 2). The obtained results provide converging evidence of some normalization in force variability (CV) but without concurrent normalization in the structure of force signal (relative power distribution, complexity). Thus, behind less variable force in the chronic stage of stroke still remain greater concentration of power in the low-frequency band (0–3 Hz) and less complex force signal (decreased SampEn) as signatures of the persisting motor impairment. The results also suggest different mechanisms underlying the recovery of force variability, force structure, and gross motor function after stroke. Force variability and structure from subacute to chronic stroke Stroke results in variable motor impairments with major recovery taking place within the first 6 months (Jorgensen et al., 1995; Gilman, 2006; Kwakkel et al., 2006). Our subjects showed improvements in all clinical outcomes but the difference did not reach statistical significance. This is likely due to a combination of factors, including a bias in recruiting less impaired subjects to comply with the task demand in the subacute phase, shallow slope of improvement from 1 to 6 months in initially less impaired subjects (Buurke et al., 2008; Verheyden

et al., 2008), ceiling effect of some of the selected clinical measures (Gladstone et al., 2002; Kwakkel et al., 2006), and a relatively small sample size. Nevertheless, our subjects attained typical recovery over the first 6 months given the initial impairment (Buurke et al., 2008; Verheyden et al., 2008). The hypothesis that the force variability, frequency composition, and complexity would normalize over the first 6 months of stroke was only supported for the force variability. Because force fluctuation is influenced by discharge variability of active motor units (Enoka et al., 2003), and the discharge patterns of motor units are disrupted after stroke due to impaired descending and afferent input to the segmental network (Dietz et al., 1986; Gemperline et al., 1995; Campanini et al., 2009), improved force variability from subacute to chronic stage may be associated with cortical reorganization during recovery from stroke (Gerloff et al., 2006). However, the recovery of force variability within the first 6 months of stroke was not associated with a broadening of the force frequency profile and greater force complexity. Computer simulations of motor-unit activity suggest that manipulation of recruitment and rate coding alone could not adequately describe the frequency composition of force fluctuation favoring instead the interaction of multiple

J. W. Chow, D. S. Stokic / Neuroscience 273 (2014) 189–198


Fig. 4. Correlations between paretic Fugl-Meyer (FM) lower extremity motor score and relative spectral power in the 0–3, 3–6, and 8–12-Hz frequency bands (left to right) for the paretic leg of stroke subjects at different force levels. The FM motor score was nonlinearly (adjungated hyperbolic or exponential growth functions) correlated with selected relative power in different frequency bands. Values given in each plot of significant correlation are the coefficient of determination (R2) and associated P-value. The horizontal dashed lines are upper (0–3-Hz band) and lower (4–6-Hz and 8–12-Hz bands) bounds of the 95% confidence intervals for control subjects.

features of motor-unit activity (Taylor et al., 2003). Thus, physiological mechanisms underlying partial normalization of the isometric force structure warrant further investigations. Force variability and structure in chronic stroke The hypothesis of a greater CV, a shift toward lower frequencies in the power spectrum, and a decrease in complexity of the force signal in the chronic stroke compared with control subjects was supported for the latter two parts only. Although the average CV across all force levels was somewhat higher in both the paretic (3.01 ± 2.70%) and non-paretic (2.03 ± 0.98%) legs compared to controls (1.98 ± 0.81%), the differences were not statistically significant (P P 0.119). These

non-significant differences in CV may be ascribed in part to the overall improved motor function after stroke, but large variances and skewness in the stroke data should also be considered, especially in the paretic leg (skewness range 2.9–4.2, mean > median). The shift in spectral power toward the low-frequency band observed in subacute stroke (Chow and Stokic, 2013) still persists in the chronic stage, but some normalization in CV likely led to disappearance of the previously reported nonlinear correlation between the CV and relative spectral power in different frequency bands. The combined findings suggest that a shift in spectral power toward lower frequencies remains a marker of disordered motor control long after stroke. Similar changes in force frequency composition have also been found in Parkinson’s disease (Vaillancourt et al., 2001) and Down


J. W. Chow, D. S. Stokic / Neuroscience 273 (2014) 189–198

syndrome (Heffernan et al., 2009). Mechanisms proposed to explain this shift include the inability to adequately control force output when processing visual information and producing motor corrections in Parkinson’s disease (Vaillancourt et al., 2001), greater closed-loop sensorimotor corrective processing in Down syndrome (Heffernan et al., 2009), and deficits in engaging additional control processes in higher frequency modulation of force in developing children (Deutsch and Newell, 2004) and older adults (Vaillancourt and Newell, 2003). It remains unknown to which extent these mechanisms explaining the downshift in force frequency apply to stroke. Our finding of decreased complexity of the force signal during isometric knee extension agrees with those reported during wrist/finger extension in chronic stroke (Lodha et al., 2010). Decreased complexity of an isometric force has also been reported in Down syndrome (Heffernan et al., 2009), Parkinson’s disease (Vaillancourt et al., 2001), and children with heavy prenatal alcohol exposure (Simmons et al., 2012). In our task, visual feedback was used to match the generated force with the target force (constantly update and reduce the difference). In the case of adequate visuomotor integration, the structure of the resulting force signal is complex and irregular, which becomes more regular if visuomotor integration is impaired (Simmons et al., 2012), as commonly seen after stroke (Rowe et al., 2009). Thus, impaired visuomotor integration may be responsible for less complex force signal after stroke, which warrants further investigations. Force structure and contraction intensity in chronic stroke Similar to the subacute stroke (Chow and Stokic, 2013), a significant shift in spectral power from the low- to high-frequency band with increasing contraction was observed in controls, but not in chronic stroke (Fig. 3d–f). Inadequate modulation of spectral power with increasing contraction has been explained by the failure to reorganize the sensorimotor output to meet the task demands (Deutsch and Newell, 2001) and by fewer degrees of freedom available to the motor system for executing motor tasks (Latash, 2012). From the physiological perspective, selective functional loss of the large, high-threshold motor units (Lukacs et al., 2008) or reduced number of recruited motor units in the paretic muscles (Li et al., 2011) may account for impaired modulation of force spectrum at stronger contractions after stroke. To gain further insights, future studies should investigate the association between cortical/ muscle activity and force signals in the frequency domain. Although our results agree with Lodha et al. (2010) in terms of a decreased complexity of force signal during an isometric task in chronic stroke, there appears to be disagreement as to how the complexity is modulated with the contraction intensity. Lodha et al. (2010) reported a steady decrease in complexity of the wrist/finger extension force from 5% to 50% MVC, but the opposite was observed here for the knee extension force. Decreased complexity of index finger and thumb forces with increasing force was also found in Parkinson’s patients (Vaillancourt et al., 2001). However, inconsistent findings

were often reported in healthy individuals (Slifkin and Newell, 1999; Sosnoff and Newell, 2007; Sosnoff et al., 2007; Prodoehl and Vaillancourt, 2010). It has been suggested that force complexity should be the greatest at moderate contraction levels (e.g., around 35% of MVC) because both motor unit recruitment and rate coding strategies could be simultaneously exploited making many degrees of freedom available for a task execution (Lodha et al., 2010). Relationship between clinical and force outcomes In addition to examining group differences between the stroke and healthy subjects, we also explored associations between the degree of clinical recovery and the force parameters in the paretic leg. The results revealed several significant nonlinear correlations between the FM lower extremity motor score and the relative spectral power in different frequency bands. These plots (Fig. 4) indicate that the stroke data points are uniformly outside the control limits across the broad range of incomplete motor recovery and that the frequency composition of the force signal tends to normalize when the gross motor recovery is nearly complete (FM score 34) . This re-affirms the notion that the shift in spectral power toward lower frequencies may be considered a marker of residual motor impairment in chronic stroke. Because our sample was biased toward higher level stroke subjects who initially required inpatient rehabilitation, future studies should explore this link further by including both more and less impaired subjects than studied here. Study limitations The results pertain to the chronic stroke population who achieved better motor recovery. Our sample was comparable to other studied chronic samples in terms of the isometric knee extension strength (Severinsen et al., 2011), FM motor scores (Wong et al., 2013; Sawacha et al., 2013), and RMI (Globas et al., 2012). Any selection bias was unintentional and introduced by the study eligibility criteria. Therefore, our results cannot be generalized to lower level stroke patients including those non-ambulatory. Although patients with clinical evidence of visual (hemianopia) and perceptual (hemineglect) deficits were excluded, they still might have some deficit in visual processing (McIntosh, 2003; Khan et al., 2008; Rowe et al., 2009). Conclusion Force variability during an isometric knee extension is not impaired among subjects who achieve relatively good motor recovery after 6 months of stroke. However, improved force variability is not accompanied by the normalization of spectral power and greater complexity of the force signal. Thus, impaired structure of the isometric force produced by knee extensors remains a marker of disordered motor control after stroke despite recovery in force variability and overall motor functions.

J. W. Chow, D. S. Stokic / Neuroscience 273 (2014) 189–198 Acknowledgments—This work was supported by the Wilson Research Foundation, Jackson, MS, USA. The authors are grateful to Mark Hemleben, Robert Hirko, Heather Maloney, Jennifer Sivak, and L. Anthony Smith for their assistance with this study.

REFERENCES Bohannon RW, Smith MB (1987) Interrater reliability of a modified Ashworth scale of muscle spasticity. Phys Ther 67:206–207. Buurke JH, Nene AV, Kwakkel G, Erren-Wolters V, Ijzerman MJ, Hermens HJ (2008) Recovery of gait after stroke: what changes? Neurorehabil Neural Repair 22:676–683. Campanini I, Merlo A, Farina D (2009) Motor unit discharge pattern and conduction velocity in patients with upper motor neuron syndrome. J Electromyogr Kinesiol 19:22–29. Chow JW, Stokic DS (2011) Force control of quadriceps muscle is bilaterally impaired in subacute stroke. J Appl Physiol 111:1290–1295. Chow JW, Stokic DS (2013) Impaired force steadiness is associated with changes in force frequency composition in subacute stroke. Neuroscience 242:69–77. Collen FM, Wade DT, Robb GF, Bradshaw CM (1991) The Rivermead Mobility Index: a further development of the Rivermead Motor Assessment. Int Disabil Stud 13:50–54. Deutsch KM, Newell KM (2001) Age differences in noise and variability of isometric force production. J Exp Child Psychol 80:392–408. Deutsch KM, Newell KM (2004) Changes in the structure of children’s isometric force variability with practice. J Exp Child Psychol 88:319–333. Dewhurst S, Graven-Nielsen T, De VG, Farina D (2007) Muscle temperature has a different effect on force fluctuations in young and older women. Clin Neurophysiol 118:762–769. Dietz V, Ketelsen UP, Berger W, Quintern J (1986) Motor unit involvement in spastic paresis. Relationship between leg muscle activation and histochemistry. J Neurol Sci 75:89–103. Freund HJ, Hefter H (1993) The role of basal ganglia in rhythmic movement. Adv Neurol 60:88–92. Enoka RM, Christou EA, Hunter SK, Kornatz KW, Semmler JG, Taylor AM, Tracy BL (2003) Mechanisms that contribute to differences in motor performance between young and old adults. J Electromyogr Kinesiol 13:1–12. Fugl-Meyer AR, Jaasko L, Leyman I, Olsson S, Steglind S (1975) The post-stroke hemiplegic patient. 1. a method for evaluation of physical performance. Scand J Rehabil Med 7:13–31. Gemperline JJ, Allen S, Walk D, Rymer WZ (1995) Characteristics of motor unit discharge in subjects with hemiparesis. Muscle Nerve 18:1101–1114. Gerloff C, Bushara K, Sailer A, Wassermann EM, Chen R, Matsuoka T, Waldvogel D, Wittenberg GF, Ishii K, Cohen LG, Hallett M (2006) Multimodal imaging of brain reorganization in motor areas of the contralesional hemisphere of well recovered patients after capsular stroke. Brain 129(Pt 3):791–808. Gilman S (2006) Time course and outcome of recovery from stroke: relevance to stem cell treatment. Exp Neurol 199:37–41. Gladstone DJ, Danells CJ, Black SE (2002) The Fugl-Meyer assessment of motor recovery after stroke: a critical review of its measurement properties. Neurorehabil Neural Repair 16:232–240. Globas C, Becker C, Cerny J, Lam JM, Lindemann U, Forrester LW, Macko RF, Luft AR (2012) Chronic stroke survivors benefit from high-intensity aerobic treadmill exercise: a randomized control trial. Neurorehabil Neural Repair 26:85–95. Heffernan KS, Sosnoff JJ, Ofori E, Jae SY, Baynard T, Collier SR, Goulopoulou S, Figueroa A, Woods JA, Pitetti KH, Fernhall B (2009) Complexity of force output during static exercise in individuals with Down syndrome. J Appl Physiol 106:1227–1233. Hsieh CL, Hsueh IP, Mao HF (2000) Validity and responsiveness of the Rivermead Mobility Index in stroke patients. Scand J Rehabil Med 32:140–142.


Jorgensen HS, Nakayama H, Raaschou HO, Vive-Larsen J, Stoier M, Olsen TS (1995) Outcome and time course of recovery in stroke. Part II: Time course of recovery. The Copenhagen Stroke Study. Arch Phys Med Rehabil 76:406–412. Khan S, Leung E, Jay WM (2008) Stroke and visual rehabilitation. Top Stroke Rehabil 15:27–36. Kouzaki M, Shinohara M, Masani K, Fukunaga T (2004) Force fluctuations are modulated by alternate muscle activity of knee extensor synergists during low-level sustained contraction. J Appl Physiol 97:2121–2131. Kwakkel G, Kollen B, Twisk J (2006) Impact of time on improvement of outcome after stroke. Stroke 37:2348–2353. Lake DE, Richman JS, Griffin MP, Moorman JR (2002) Sample entropy analysis of neonatal heart rate variability. Am J Physiol Regul Integr Comp Physiol 283:R789–R797. Latash ML (2012) The bliss (not the problem) of motor abundance (not redundancy). Exp Brain Res 217:1–5. Li X, Wang YC, Suresh NL, Rymer WZ, Zhou P (2011) Motor unit number reductions in paretic muscles of stroke survivors. IEEE Trans Inf Technol Biomed 15:505–512. Lodha N, Naik SK, Coombes SA, Cauraugh JH (2010) Force control and degree of motor impairments in chronic stroke. Clin Neurophysiol 121:1952–1961. Lodha N, Coombes SA, Cauraugh JH (2012) Bimanual isometric force control: asymmetry and coordination evidence post stroke. Clin Neurophysiol 123:787–795. Lodha N, Misra G, Coombes SA, Christou EA, Cauraugh JH (2013) Increased force variability in chronic stroke: contributions of force modulation below 1 Hz. PLoS One 8:e83468. Lukacs M, Vecsei L, Beniczky S (2008) Large motor units are selectively affected following a stroke. Clin Neurophysiol 119:2555–2558. Marsden CD (1978) The mechanisms of physiological tremor and their significance for pathological tremors. In: Desmedt JE, editor. Progress in clinical neurophysiology, Vol. 5, physiological tremor, pathological tremor and clonus. Basel: Karger. p. 1–16. McAuley JH, Marsden CD (2000) Physiological and pathological tremors and rhythmic central motor control. Brain 123(Pt 8):1545–1567. McIntosh C (2003) Stroke revisited: visual problems following stroke and their effect on rehabilitation. Br Orthopt J 60:10–14. Muceli S, Farina D, Kirkesola G, Katch F, Falla D (2011) Reduced force steadiness in women with neck pain and the effect of short term vibration. J Electromyogr Kinesiol 21:283–290. Pincus SM (1991) Approximate entropy as a measure of system complexity. Proc Natl Acad Sci U S A 88:2297–2301. Prodoehl J, Vaillancourt DE (2010) Effects of visual gain on force control at the elbow and ankle. Exp Brain Res 200:67–79. Richman JS, Moorman JR (2000) Physiological time-series analysis using approximate entropy and sample entropy. Am J Physiol Heart Circ Physiol 278:H2039–H2049. Rowe F, Brand D, Jackson CA, Price A, Walker L, Harrison S, Eccleston C, Scott C, Akerman N, Dodridge C, Howard C, Shipman T, Sperring U, MacDiarmid S, Freeman C (2009) Visual impairment following stroke: do stroke patients require vision assessment? Age Ageing 38:188–193. Sawacha Z, Carraro E, Contessa P, Guiotto A, Masiero S, Cobelli C (2013) Relationship between clinical and instrumental balance assessments in chronic post-stroke hemiparesis subjects. J Neuroeng Rehabil 10:95. Schiffman JM, Luchies CW, Piscitelle L, Hasselquist L, Gregorczyk KN (2006) Discrete bandwidth visual feedback increases structure of output as compared to continuous visual feedback in isometric force control tasks. Clin Biomech (Bristol, Avon) 21:1042–1050. Severinsen K, Jakobsen JK, Overgaard K, Andersen H (2011) Normalized muscle strength, aerobic capacity, and walking performance in chronic stroke: a population-based study on the potential for endurance and resistance training. Arch Phys Med Rehabil 92:1663–1668.


J. W. Chow, D. S. Stokic / Neuroscience 273 (2014) 189–198

Simmons RW, Nguyen TT, Levy SS, Thomas JD, Mattson SN, Riley EP (2012) Children with heavy prenatal alcohol exposure exhibit deficits when regulating isometric force. Alcohol Clin Exp Res 36:302–309. Slifkin AB, Newell KM (1999) Noise, information transmission, and force variability. J Exp Psychol Hum Percept Perform 25:837–851. Slifkin AB, Vaillancourt DE, Newell KM (2000) Intermittency in the control of continuous force production. J Neurophysiol 84:1708–1718. Sosnoff JJ, Newell KM (2007) Are visual feedback delays responsible for aging-related increases in force variability? Exp Aging Res 33:399–415. Sosnoff JJ, Deutsch KM, Newell KM (2007) Does muscular weakness account for younger children’s enhanced force variability? Dev Psychobiol 49:399–405.

Taylor AM, Christou EA, Enoka RM (2003) Multiple features of motorunit activity influence force fluctuations during isometric contractions. J Neurophysiol 90:1350–1361. Vaillancourt DE, Newell KM (2003) Aging and the time and frequency structure of force output variability. J Appl Physiol 94:903–912. Vaillancourt DE, Slifkin AB, Newell KM (2001) Intermittency in the visual control of force in Parkinson’s disease. Exp Brain Res 138:118–127. Verheyden G, Nieuwboer A, De Wit L, Thijs V, Dobbelaere J, Devos H, Severijns D, Vanbeveren S, De Weerdt W (2008) Time course of trunk, arm, leg, and functional recovery after ischemic stroke. Neurorehabil Neural Repair 22:173–179. Wong SS, Yam MS, Ng SS (2013) The Figure-of-Eight Walk test: reliability and associations with stroke-specific impairments. Disabil Rehabil 35:1896–1902.

(Accepted 10 May 2014) (Available online 16 May 2014)

Variability, frequency composition, and complexity of submaximal isometric knee extension force from subacute to chronic stroke.

We examined changes in the variability, frequency composition, and complexity of force signal from subacute to chronic stage of stroke during maintena...
2MB Sizes 1 Downloads 3 Views