Novel kinematic indices for quantifying movement agility and smoothness after cervical Spinal Cord Injury

Abstract

BACKGROUND:

After cervical Spinal Cord Injury (SCI), upper limb movements made by patients have a lack of smoothness and a hand velocity profile characterized by a high number of velocity peaks.

OBJECTIVE:

The aim of the present paper is to propose three novel kinematic indices for quantifying movement agility and smoothness, and to analyze their discriminative capability between healthy and pathological people.

METHODS:

18 people, healthy and two groups of patients with cervical SCI, participated in the study. Kinematic indices in relation to movement agility and smoothness were computed from hand trajectories and velocity profiles during the performance of the ADL of drinking from a glass.

RESULTS:

The proposed indices discriminated between healthy and SCI people. The results are greater in healthy than SCI people. Both smoothness indices detected significant differences between healthy and both SCI groups. Moreover, the Agility index showed capacity for discriminating between both patients groups.

CONCLUSIONS:

The main contribution of this research consists on the proposal of kinematic indices from experimental data, whose results are dimensionless and relative to a pattern of healthy subjects. We hope that kinematic indices proposed are a step toward the standardization of the quantitative assessment of movement characteristics and functional impairments.

Keywords

Kinematic indices functional assessment upper limb activities of daily living neurological diseases

Abbreviations

SCI

Spinal Cord Injury

ADL

Activity of Daily Living

Upper Limb

1 Introduction

The incidence of Spinal Cord Injury (SCI) varies greatly worldwide from 12.1 to 57.8 SCI cases per million population depending on the countries (Van den Berg et al., 2010). Among them, more than the 50% of people with SCI have upper limb (UL) motor function impairments, resulting in limitations in performing a variety of functional tasks. In that condition, primary objectives of the rehabilitation effort are to maximize the available muscle performance and teach new skills in order for the patient to achieve full functional capabilities (Beninatoet al., 2004; Schmitz, 2001). Specifically, smoothness is a characteristic of unimpaired and well-coordinated movements (Rohrer et al., 2002). In presence of movement disorders, the hand velocity profile is impaired (de los Reyes-Guzmán et al., 2010; Van Der Heide et al., 2005) and as a consequence, the movement characteristics in relation to the agility and smoothness are impaired too.

Kinematic studies have been introduced in clinical settings to perform a biomechanical characterization of several tasks and Activities of Daily Living (ADL) in an objective way, in healthy people (Aizawa et al., 2010; Magermans et al., 2005; Murgia et al., 2010; Murphy et al., 2006; Petuskey et al., 2007; van Andel et al., 2008) and after neurological diseases such as stroke (Kim et al., 2014; Lang et al., 2005; Murphy et al., 2011, 2013; Osu et al., 2011), cerebral palsy (Butler et al., 2010; Jaspers et al., 2011; Klotz et al., 2013) or SCI (de los Reyes-Guzmán et al., 2010). 3D movement analysis equipments record large data sets depending on the sample frequency used. The efficient management and processing of raw kinematic data is necessary to detect changes in patients’ UL functionality.

Indices about movement characteristics from kinematic data require a precise mathematical and numerical method for their quantification. Smooth, well-coordinated movements are a characteristic of a healthy human motor behaviour. Similarly, subjects affected by neurological diseases have UL movements composed by a large number of submovements with loose temporal packing (Rohrer et al., 2002). Kinematic indices would enable to analyze learning, motor recovery and assess long-term improvements after neurological diseases.

In order for a movement characteristic measure to be useful, it must satisfy the following requirements (Balasubramanian et al., 2012): i) It must be dimensionless, independent of the signal amplitude and duration; ii) It must have a monotonic response to the motion characteristics so that an increase in the kinematic index corresponds to an increase in the movement characteristic; iii) It must be sensitive to changes in movement characteristics within the physiological ranges, and iv) It must be computationally inexpensive and robust to the noise.

However, it has been shown that existing indices do not satisfy these requirements (de los Reyeset al., 2014). Most of these measurements are usually expressed as absolute values in the corresponding international system units (Butler et al., 2010; Chang et al., 2005; Colombo et al., 2008; Dipietro et al., 2007; Merlo et al., 2012; Murphy et al., 2006; Rohrer et al., 2002; Van Der Heide et al., 2005) and, sometimes an increase in the kinematic index corresponds to a decrease in the movement characteristic that represents (Colombo et al., 2008). Other aspect to take into account is that the UL movement most commonly analyzed is the reaching and point-to-point movements (Celik et al., 2010; Chang et al., 2005; Colombo et al., 2008; Jaspers et al., 2011, Kamperet al., 2002; Lang et al., 2005, 2006; Merlo et al., 2012; Rohrer et al., 2002; Wagner et al., 2008). However, this research is based on the assessment of a complete ADL.

In this context, the objectives of the present paper are to define kinematic indices for quantifying movement agility and smoothness, during the performance of the ADL of drinking, and to check the capability of the proposed indices for discriminating between healthy and people with cervical SCI. The main contribution is that indices are defined from experimental data and results are expressed dimensionless and relative to a pattern of healthy subjects.

2 Methods

2.1 Participants

A total of 18 people divided into three groups participated in the study: a healthy group (n = 7); and two groups of patients with cervical SCI with metameric level C6 (n = 7) and C7 (n = 4). This patients’ sample was chosen because the two groups have UL functional differences between them. Hence, patients that were included into the C6 group are more affected that those in the other one. C6 patients retain control of the elbow flexion and the wrist extensor muscles but lose active extension of the elbow, whereas C7 patients retain active control of all these muscles. All participants were right handed. Background data of participants are provided in Table 1. The patients screened had to fulfill the following criteria to be included in the study: age 16 to 65 years, injuryof at least 6 months’ duration and level of injury C6 or C7 classified according to the American Spinal Injury Association (ASIA) (Maynard et al., 1997) scale into grades A or B. Patients who presented any vertebral deformity, joint constraint, surgery or any of the UL, balance disorders, dysmetria due to associated neurologic disorders, visual acuity defects, cognitive deficit, or head injury associated with the SCI were excluded. Patients were classified into C6 and C7 SCI by a physical examination. The UL Motor Index was obtained (Maynard et al., 1997), with the assessment of the strength of five muscles groups of the right UL by a physiotherapist. Each muscle group can be assessed between 0 (no function) to 5 (normal function) with a total of 25 points. The guidelines of the declaration of Helsinki were followed in every case. Informed consent was obtained from all individual participants included in the study, which was approved by the Local Ethics Committee, Toledo, Spain.

2.2 Procedure

UL movement was recorded by using the Codamotion system (Charnwood Dynamics, Ltd, UK) based on active markers. A total of 18 markers were used, placed on the skin surface in the trunk and the right arm (Fig. 1). All the participants, instrumented with Codamotion markers and seated in a wheelchair in front of a table, performed only one experimental session with 5 repetitions of the ADL of drinking. For computing kinematic data from markers data position, a biomechanical model was developed. The model included the trunk, right arm, right forearm and the right hand body segments. All this experimental setup was described in detail in a previous study (de los Reyes-Guzmán et al., 2010). The only difference within the experimental protocol was the distance to the glass placed on the table. In this study the glass was placed in the midline of the body, to the 75% of the maximal UL reaching (Butler et al., 2010) with the aim of minimizing compensatory movements.

The drinking task included reaching and grasping the glass, lifting the glass to the mouth, drinking a swallow, releasing the glass on the table and returning to the starting position. All the participants performed the movement with the right arm. They were instructed for initiate the drinking task at a comfortable self-selected speed. Five trials of the task were recorded for processing. The mean value for each variable was used for statistical analysis.

3D motion analysis was performed with Codamotion (Charnwood Dynamics, Ltd, UK) photogrammetry system. 3D marker positions are calculated instantly with a spatial resolution of 0.1 mm. Position data were exported to Visual3D for calculating kinematic data. Data were filtered with second-order Butterworth filter with a cutoff frequency of 4 Hz. The new contribution was based on exporting kinematic data to MATLAB (The MathWorks, Natick, USA) software for calculating the kinematic indices.

To facilitate analysis, the drinking ADL was broken down into 5 consecutive phases, following Murphy’s study guidelines (Murphy et al., 2006): reaching (included grasping the glass), forward transport, drinking, distal transport (included releasing the object) and returning to the starting point.

2.3 Kinematic indices

Three kinematic indices were defined. One of them assesses UL movement agility and two indices movement smoothness. Motor learning is fundamental to neurological rehabilitation. UL functional deficits after neurological diseases are reflected in compensatory movements in proximal joints. Hence, these indices were designed to detect UL functional impairments.

The proposal of these novel kinematic indices arised from meetings with amultidisciplinar group composed by people with engineering and clinicalprofiles. The three indices presented in this paper werecomputing from the hand velocity profile during the movement. The hand velocity was obtained as the first derivative from the hand trajectory performed and was applied as the module of the hand velocity vector obtained (Equation 1). $v [n] = \sqrt{v_{x} [n]^{2} + v_{y} [n]^{2} + v_{z} [n]^{2}}$ (1)

The velocity profiles of a healthy subject and a patient with cervical SCI, randomly selected, are shown in Fig. 2. The healthy and pathological velocity profiles present some differences that suggest the proposal of these indices from these kinematic data. In the healthy profile, four predominant velocity peaks are observed. Each of them corresponds to a phase within the ADL of drinking in which there is a displacement of the hand in the space: reaching, forward transport, distal transport and returning phases (Fig. 2).

2.3.1 Agility index

Agility (Ag) could be defined as the capacity for executing a movement in a fast way. But by means of meetings maintained with clinical staff, in the rehabilitation context, it wasn’t enough. It was necessary that the movement execution by a patient could include requirements of control and accuracy. Hence, the proposal of the agility index took into account parameters in relation to the movement velocity and the accuracy index (De los Reyes-Guzmán et al., 2014), applied in this research during the complete cycle corresponding to the ADL of drinking.

So the mathematical formulation proposed took into account velocity and accuracy requirements, computing the index as the product of three terms (Equation 2):

$A_{g} = γ μ A$ (2)

The γ parameter was a dependent function on the movement time (γ=γ(t)); μ parameter was defined as a dependent function on the difference between the maximal and the mean velocity (v_max and v_mean, respectively) computed during the complete cycle of the ADL of drinking (Equation 3). So, v_dif =Δv and μ is function of Δv (μ(Δv)). $v_{dif} = v_{max} - v_{mean}$ (3)

The result of the agility index was normalized as the reference healthy pattern (A_{_gref}) (Equation 4). $A_{gnorm} = \frac{A_{g}}{A_{g_{ref}}} 100$ (4)

γ was an adimensionless parameter that contributes in a positive or negative way to the Agility index depending on its value is greater or lower than 1. The possible interval values were between 0 and 1.2, and the γ value decreased as the movement time increased. The value γ= 1 was obtained when the subject performed the complete ADL cycle with a movement time equal to the mean time movement obtained by the healthy pattern. If a person executed the activity in a minor time, the γ value would be greater than 1. In this case, the γ parameter had a positive contribution to the Agility index. The γ function was defined in a complete way by choosing the value of the constant β₂ (Equation 5, Fig. 3).

$γ (t) = 1 + β_{2} (1 - \frac{t}{t_{ref}})$ (5)

The constant β₂ = 1/5 was chosen, because the maximal movement time of aproximately 40 s was enough for pathological people could execute a complete cycle of the drinking task.

μ was an adimensionless parameter, dependent on the v_dif value that contributed in a negative way to the Agility index. From experimental data, the mean value of v_dif was 0.65 m/s for the healthy pattern. Patients with cervical SCI usually performed UL abrupt movements which produce higher velocity peaks. However, they needed more time for executing the ADL of drinking and, as consequence, the movement mean velocity was lower than in healthy people, due to longer valleys between consecutive velocity peaks (Rohrer et al., 2002). So the more difference between the maximal and mean velocities, a lower value of the μ parameter (Fig. 4), and lower value of the global Agility index.

But μ parameter was completely defined by choosing the value of the constant β₃ (Equation 6). In this case, β₃ had been fixed to 1/20, because v_dif ∼1.5 m/s hadn’t been reached from experimentaldata.

$μ (v_{dif}) = {\begin{matrix} 1, & v_{dif} \leq v_{{dif}_{ref}} \\ 1 + β_{3} (1 - \frac{v_{dif}}{v_{{dif}_{ref}}}), & v_{dif} \geq v_{{dif}_{ref}} \end{matrix}$ (6)

2.3.2 Smoothness indices

The objective of the smoothness indices was to assess the good-coordination level in the UL movements that allowed discriminate between healthy and pathological people.

In this research, two indices were proposed for assessing smoothness. The first one, S₁, was based on computing the peaks number (N_peaks) from the hand velocity profile during the complete movement. The other one, S₂, was computed from the Fourier Transform applied to the hand velocity profile.

Smoothness index from the peaks number (S₁)

The Smoothness index S₁ computed the peaks number from the velocity profile module vector during the complete cycle of the ADL of drinking. Each velocity peak corresponded to an acceleration and deceleration period (Chang et al., 2005; Fasoli et al., 2002; Merlo et al., 2012; Rohrer et al., 2002). The time between two consecutive velocity peaks had to be at least 150 ms (Murphy et al., 2011), and a peak has to be greater than the 10% of the global peak during the complete cycle of the ADL. This method minimized the noise effect in the signal.

A greater peaks number indicated a more fragmented movement and, as consequence, a less smooth movement. Moreover, this kinematic index was normalized as the mean peaks number for the healthy pattern, N_peaksref (Equation 7). $S_{1} = \frac{N_{{peaks}_{ref}}}{N_{peaks}} 100$ (7)

Smoothness index from the Fourier analysis (S₂)

The smoothness index S₂ was computed from velocity profile’s Fourier magnitude spectrum during the complete cycle of the drinking task. In the practice, the frequency spectrum was obtained from the DFT (Discrete Fourier Transform), a sampled version of the DTFT (Discrete-Time Fourier Transform) in which n = 0 and n = N-1 correspond to the starting and ending movement events (Equation 8).

$\begin{matrix} V [k] = \sum_{n = 0}^{N - 1} v [n] e^{- j \frac{2 π kn}{N}}, \\ n \in [0, N - 1] y k \in [0, N - 1] \end{matrix}$ (8)

Matlab software computes the DFT by means of the FFT (Fast Fourier Transform), reducing the computational cost due to a minor operations number. So the spectral arc-length (l_spect) was computed by the equation proposed by Balasubramanian. In this research, this equation had been extended with an aditional factor, whose value depended on the maximal frequency to which the spectrum reached (Equation 9). The cut-off frequency, fc, value had been fixed to 20 Hz (w = 40π rad), because this frequency contained all the signal information in relation to frequency. The Kc constant corresponded to the Fourier spectrum sample of the maximal frequency to wich the spectrum reached (Fig. 5). $\begin{matrix} l_{spect} = \frac{f_{c}}{f_{c} - f_{max}} \\ \sum_{k = 1}^{K_{c} - 1} \sqrt{{(\frac{1}{K_{c} - 1})}^{2} + {(Δ \hat{V} [k])}^{2}} \end{matrix}$ (9) $Δ \ddot{V} [k] = \ddot{V} [k] - \ddot{V} [k - 1], k \in [1, K - 1]$

Finally, the smoothness index S₂ was computed normalizing the spectral arc-length by the mean spectral arc-length for the healthy pattern (Equation 10). $S_{2} = \frac{l_{{spect}_{ref}}}{l_{spect}} 100$ (10)

2.4 Statistical analysis

Statistical analysis was performed with SPSS (Statistical Packages for Social Sciences, release 12.0 for Windows, SPSS Inc, Chicago, IL). In the analysis of kinematic indices, the mean value of the five recordings was used.

A descriptive analysis was made of the clinical and functional variables by calculating the median and interquartile range of the quantitative variable and the frequencies and percentages of the qualitative variables.

To check the discriminative capability of kinematic indices proposed, comparisons between healthy and SCI patients, and between patients with different severity level were made. The Kruskal-Wallis test was applied to find possible differences in each variable between the three groups analyzed; the Kruskal-Wallis test is p < 0.05, the equivalence of behavior between groups can be rejected and a pairwise comparison can be made using the U Mann-Whitney test. The Bonferroni correction was applied, which takes into account randomness due to multiple comparisons.

The repeatability of the experimental protocol during the ADL of drinking was analyzed in a previous study (de los Reyes-Guzmán et al., 2010).

3 Results

The sample analyzed was broken down into three groups that were matched in age, weight and height. So no statistical significant differences were found between the three groups.

3.1 Agility index

The agility index detected differences between the three populations analyzed. Agility was greater in healthy subjects than in C6 and C7 SCI groups (p < 0.05 and p < 0.01, respectively) during the complete ADL. Moreover, this index discriminated between injury levels C6 and C7 (p < 0.05). The C6 SCI group was functionally more affected than C7 group and the accuracy result is lower in C6 than in C7 SCI group (Table 2). The boxplot diagrams weren’t overlapped between the three groups analyzed (Fig. 6).

μ and γ parameters were able to discriminate between healthy and C6 SCI people (p < 0.05), but the discriminating character of the agility index was coming for the Accuracy index, previously calaculated as the deviation between the hand trajectory of the patient and the mean trajectory for the healthy pattern (De los Reyes-Guzmán et al., 2014).

The μ parameter was a dependent function on the difference between the maximal and mean velocities during the movement analyzed. The maximal velocity was significantly greater in C6 SCI people (0.98 m/s) than in healthy pattern (0.86 m/s) (p < 0.05). However, the mean velocity during the movement was greater in healthy people than in C6 SCI group. So the significant difference found between healthy and C6 SCI group in the μ parameter.

3.2 Smoothness indices

Both smoothness indices, S₁ y S₂, were statistically different between the healthy subjects and both groups of SCI people (p < 0.01). Smoothness is a characteristic of well-coordinated movements and the index S₁ was greater in healthy people (100.20 ± 6.79 %) than in C6 (54.91 ± 12.17) and C7 SCI people (60.53 ± 15.24) (Table 3, Fig. 6). This index was calculated from the peaks number in the velocity profile. So, the peaks number during the complete ADL was greater in both groups of SCI patients (12.45 and 13.70 peaks, respectively) than in healthy (6.97 peaks) (p < 0.01). However, in the reaching phase, the peaks number was statistically different only between healthy and C6 SCI patients (p < 0.05).

In relation to the smoothness index S₂, the same differences were found between the three groups, but in this case, the results were aproximately equal in both groups of patients. However, the intercuartile range was greater in C6 (14.01%) than in C7 (6.11%) SCI patients.

4 Discussion

In this paper, three novel kinematic indices for quantifying movement agility and smoothness, during the performance of the ADL of drinking have been defined. Then, the capability of these indices for discriminating between healthy and people with UL motor impairments has been checked. The results show that the proposed indices are adequated for detecting functional UL impairments, when they have been applied to people with cervical SCI. The main contribution is that indices results are dimensionless and relative to a pattern of healthy subjects. This is a limitation noticed in previous studies, in which the indices results are expressed as absolute values in the corresponding units depending on the variable analyzed (Chang et al., 2005; de los Reyes-Guzmánet al., 2010; Murphy et al., 2006; Rohrer et al., 2002). Other important contribution of this research is the UL assessment during the performance of complete ADL, while the most research articles are centered in the reaching and point-to-point movement analysis (Bosecker et al., 2010; Celik et al., 2010; Chang et al., 2005; Colombo et al., 2008; Kamper et al., 2002; Lang et al., 2006; Merlo et al., 2012; Rohrer et al., 2002). The ADL of drinking from a glass has been selected as a representative ADL because its correct execution for maintaining the glass in the hand, requires UL control, coordination, accuracy and force. These characteristics and motor control aspects are impaired in presence of neurological diseases. For that reason, although the kinematic indices have been applied to patients with cervical SCI, this methodology could be applied to people with other neurological pathologies that produce upper limb movement disorders such as stroke or cerebral palsy.

The kinematic indices presented in this research, have been computed from the velocity profile of the hand movement during the performance of the drinking ADL. Until now, no evidence has been found in relation to kinematic indices, such as those proposed in this research, with the aim of assessing a complete ADL.

In literature, kinematic indices like the Agility index proposed have not been found. This kinematic index has been defined from hand movement velocity data for providing information about the velocity and velocity changes during the movement. However, from meetings with clinical staff, this isn’t enough and accuracy requirements are neccesary. From the observed differences in these velocity and acceleration parameters, the Agility index has shown capacity for discriminating between the three populations analyzed. The healthy people that compose the healthy pattern performed the ADL of drinking in a mean of 6.76 s. This result is corresponded to a similar previous study following the same experimental protocol (6.49 s) (Murphy et al., 2011). In this research, it has been noticed that SCI people require a longer time for performing a complete cycle of the drinking task. The mean velocity, v_mean, during the complete ADL has been lower in C6 and C7 SCI people than in healthy subjects, but there are no significant differences between the three groups analyzed. However, the peak velocity, v_max, was greater in both SCI groups than in healthy people due to the performance of abrupt UL movements. Moreover, this index was computed by the contribution of the μ parameter, as a measurement of the difference between the v_max and v_mean. This difference is greater and statistically significant in C6 SCI people than in healthy. However, in people who have suffered stroke, the opposite effect has been noticed and the v_max was lower than in healthy during the reaching phase within the ADL of drinking (Murphy et al., 2011).

Smoothness and well-coordinated movements is a characteristic of a healthy pattern and well-performed movements (Balasubramanian et al., 2012). Smoothness metric has been found in literature applied to UL reaching movements, and computed from the peaks number in the velocity curve, called movement units (Murphy et al., 2011). Other studies have proposed this metric as the ratio between the maximum and the mean velocity during the movement (v_max/v_med) (Bosecker et al., 2010; Merlo et al., 2012; Rohreret al., 2002; Zariffa et al., 2012), or the jerk metric as a measurement of non-smoothness (Hogan et al., 2009; Rohrer et al., 2002). The smoothness quantification in terms of the ratio between both velocities hasn’t been analyzed in this research because the Agility index proposed takes into account the differences between maximal and mean velocity during the movement. Several metrics have been proposed by Balasubramanian with the aim of quantifying the movement smoothness in people with stroke. This author found smoothness metric from the peaks number is very sensible to the noise, proposing other alternative metrics, some of them based on the application of the Fourier Transform (Balasubramanianet al., 2012).

So in this research two kinematic indices have been proposed for assessing movement smoothness during the complete cycle of the ADL of drinking. The first proposal is computed from the peaks number in the velocity profile of the hand movement, and the second one computed from the Fourier magnitude spectrum of the velocity profile. Both smoothness indices are discriminative between the healthy and cervical SCI people. But they haven’t ability for discriminating between C6 and C7 tetraplegia levels.

In relation to the smoothness index S₁, the velocity peaks number has been computed during the complete drinking task. Each velocity peak corresponds to an acceleration and deceleration period (Chang et al., 2005; Fasoli et al., 2002; Merlo et al., 2012; Rohrer et al., 2002). Previous studies have analyzed the velocity profile of this ADL (Butler et al., 2010; Murphy et al., 2011, 2013). Murphy computed the peaks number during the reaching and forward transport phases within the drinking task. In this situation the theoretical peaks number is two, one in each movement phase. The velocity profile in each phase analyzed is continuous, smooth, bell-shaped, and a predominant peak (Murphy et al., 2011) when the movement is performance by healthy people. However, in people with UL movement disorders perform, the movements analyzed have a higher number of velocity peaks. Murphy, in her study, showed that people with stroke performed the reaching and forward transport phases with 8.4 ± 4.2 peaks, whereas healthy people showed 2.3 ± 0.3 peaks, detecting statistical signification between healthy and stroke people (Murphy et al., 2011). In other study, for the healthy subjects group 4.8 ± 1.2 peaks were obtained, and 17.5 ± 4.8 peaks for hemiplegic people, detecting significant differences between them (Butler et al., 2010). In this research, statistical differences have been found between the healthy group (6.97 ± 0.53 peaks) and both SCI people groups, C6 SCI (12.45 ± 2.21) and C7 SCI (13.70 ± 4.63). A greater peaks number indicates a more fragmented movement and, as consequence, a less smooth movement. For that reason, authors have shown the results by a negative sign. In this way, an increases in the metric correspond to less peaks in the velocity profile (Colombo et al., 2008; Balasubramanian et al., 2012a). In this research, the index based on the peaks number has been normalized by the mean peaks number for the healthy subjects that compose the reference pattern. The global result is a smoothness index S₁ with capacity for discriminating between healthy and SCIpeople.

Finally, with the smoothness index S₂ a frequency spectral analysis was made during the complete drinking task. Balasubramaniam proposes this index as a smoothness measure with the aim of discriminating between several affectation levels of stroke, under the initial hypothesis that an irregular velocity profile with a greater peaks number, may present a more complex spectrum (Balasubramanian et al., 2012). In this research the results of smoothness index S₂ have been shown normalized by the healthy pattern and the index shows capacity for distinguising between healthy and C6 and C7 SCIpeople.

In this study, the discriminative capability of the indices proposed has been proved. However, this study has limitations. For using this UL assessment instrument within a clinical setting, the validity, reproducibility and the responsiveness to the change of the kinematic indices have to be analyzed.

5 Conclusions

In this paper, three kinematic indices have been proposed with the aim of quantifying UL functional impairments in people with SCI. These indices, agility and two proposals for smoothness indices are computed from the velocity profile of the hand movement during the performance of the ADL of drinking. They are adequated for discriminating between healthy and people with different metameric level of cervical SCI.

Conflict of interest

The authors have no conflict of interest to declare.

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and local research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Footnotes

Acknowledgments

The research for this manuscript has been partially funded by grant from the Spanish Ministry of Science and Innovation CONSOLIDER INGENIO, project HYPER (Hybrid NeuroProsthetic and neuroRobotic Devices for Functional Compensation and Rehabilitation of Motor Disorders, CSD 2009-00067).

References

Aizawa

, Masuda

, Koyama

, Nakamaru

, Isozaki

, Okawa

, & Morita

(2010). Three-dimensional motion of the upper extremity joints during various activities of daily living. Journal of biomechanics, 43(15), 2915–2922.

Balasubramanian

, Melendez-Calderon

, & Burdet

(2012). A robust and sensitive metric for quantifying movement smoothness. Biomedical Engineering, IEEE Transactions on, 59(8), 2126–2136.

Balasubramanian

, Colombo

, Sterpi

, Sanguineti

, & Burdet

(2012a) Robotic assessment of upper limb motor function after stroke. American Journal of Physical Medicine & Rehabilitation, 91(11), S255–S269.

Beninato

, O’Kane

K. S.

, & Sullivan

P. E.

(2004). Relationship between motor FIM and muscle strength in lower cervical-level spinal cord injuries. Spinal Cord, 42(9), 533–540.

Bosecker

, Dipietro

, Volpe

, & Krebs

H. I.

(2010). Kinematic robot-based evaluation scales and clinical counterparts to measure upper limb motor performance in patients with chronic stroke. Neurorehabilitation and Neural Repair, 24(1), 62–69.

Butler

E. E.

, Ladd

A. L.

, LaMont

L. E.

, & Rose

(2010). Temporal–spatial parameters of the upper limb during a reach & grasp cycle for children. Posture, 32(3), 301–306.

Cacho

E. W. A.

, de Oliveira

, Ortolan

R. L.

, Varoto

, & Cliquet

(2011). Upper limb assessment in tetraplegia: Clinical, functional and kinematic correlations. International Journal of Rehabilitation Research, 34(1), 65–72.

Celik

, O’Malley

M. K.

, Boake

, Levin

H. S.

, Yozbatiran

, & Reistetter

(2010). Normalized movement quality measures for therapeutic robots strongly correlate with clinical motor impairment measures. Neural Systems and Rehabilitation Engineering, IEEE Transactions on, 18(4), 433–444.

Chang

J. J.

, Wu

T. I.

, Wu

W. L.

, & Su

F. C.

(2005). Kinematical measure for spastic reaching in children with cerebral palsy. Clinical Biomechanics, 20(4), 381–388.

10.

Colombo

, Pisano

, Micera

, Mazzone

, Delconte

, Carrozza

M. C.

, & Minuco

(2008). Assessing mechanisms of recovery during robot-aided neurorehabilitation of the upper limb. Neurorehabilitation and Neural Repair, 22(1), 50–63.

11.

de los Reyes-Guzmán

, Dimbwadyo-Terrer

, Trincado-Alonso

, Monasterio-Huelin

, Torricelli

, & Gil-Agudo

(2014). Quantitative assessment based on kinematic measures of functional impairments during upper extremity movements: A review. Clinical Biomechanics, 29(7), 719–727.

12.

De los Reyes-Guzmán

, Dimbwadyo-Terrer

, Pérez-Nombela

, Trincado

, Torricelli

, & Gil-Agudo

(2014). Objective metrics for functional evaluation of upper limb during the ADL of drinking: Application in SCI. In Springer International Publishing. XIII Mediterranean Conference on Medical and Biological Engineering and Computing, 1751–1754.

13.

de los Reyes-Guzmán

, Gil-Agudo

, Peñasco-Martín

, Solís-Mozos

, del Ama-Espinosa

, & Pérez-Rizo

(2010). Kinematic analysis of the daily activity of drinking from a glass in a population with cervical spinal cord injury. Journal of Neuroengineering and Rehabilitation, 7(1), 41.

14.

Dipietro

, Krebs

H. I.

, Fasoli

S. E.

, Volpe

B. T.

, Stein

, Bever

, & Hogan

(2007). Changing motor synergies in chronic stroke. Journal of Neurophysiology, 98(2), 757–768.

15.

Fasoli

S. E.

, Trombly

C. A.

, Tickle-Degnen

, & Verfaellie

M. H.

(2002). Effect of instructions on functional reach in persons with and without cerebrovascular accident. American Journal of Occupational Therapy, 56(4), 380–390.

16.

Hogan

, & Sternad

(2009). Sensitivity of smoothness measures to movement duration, amplitude, and arrests. Journal of Motor Behavior, 41(6), 529–534.

17.

Jaspers

, Desloovere

, Bruyninckx

, Klingels

, Molenaers

, Aertbeliën

, & Feys

(2011). Three-dimensional upper limb movement characteristics in children with hemiplegic cerebral palsy and typically developing children. Research in Developmental Disabilities, 32(6), 2283–2294.

18.

Kamper

D. G.

, McKenna-Cole

A. N.

, Kahn

L. E.

, & Reinkensmeyer

D. J.

(2002). Alterations in reaching after stroke and their relation to movement direction and impairment severity. Archives of Physical Medicine and Rehabilitation, 83(5), 702–707.

19.

Kim

, Song

W. K.

, Lee

H. Y.

, Park

D. S.

, Ko

B. W.

, & Kim

(2014). Kinematic analysis of upper extremity movement during drinking in hemiplegic subjects. Clinical Biomechanics, 29(3), 248–256.

20.

Klotz

M. C. M.

, Kost

, Braatz

, Ewerbeck

, Heitzmann

, Gantz

, & Wolf

S. I.

(2013). Motion capture of the upper extremity during activities of daily living in patients with spastic hemiplegic cerebral palsy. Posture, 38(1), 148–152.

21.

Lang

C. E.

, Wagner

J. M.

, Edwards

D. F.

, Sahrmann

S. A.

, & Dromerick

A. W.

(2006). Recovery of grasp versus reach in people with hemiparesis poststroke. Neurorehabilitation and Neural Repair, 20(4), 444–454.

22.

Lang

C. E.

, Wagner

J. M.

, Bastian

A. J.

, Hu

, Edwards

D. F.

, Sahrmann

S. A.

, & Dromerick

A. W.

(2005). Deficits in grasp versus reach during acute hemiparesis. Experimental Brain Research, 166(1), 126–136.

23.

Magermans

D. J.

, Chadwick

E. K. J.

, Veeger

H. E. J.

, & Van Der Helm

F. C. T.

(2005). Requirements for upper extremity motions during activities of daily living. Clinical Biomechanics, 20(6), 591–599.

24.

Maynard

F. M.

, Bracken

M. B.

, Creasey

G. J. F. D.

, Ditunno

J. F.

, Donovan

W. H.

, Ducker

T. B.

, & Young

(1997). International standards for neurological and functional classification of spinal cord injury. Spinal Cord, 35(5), 266–274.

25.

Merlo

, Longhi

, Giannotti

, Prati

, Giacobbi

, Ruscelli

, & Mazzoli

(2012). Upper limb evaluation with robotic exoskeleton. Normative values for indices of accuracy, speed and smoothness. NeuroRehabilitation, 33(4), 523–530.

26.

Murgia

, Kyberd

, & Barnhill

(2010). The use of kinematic and parametric information to highlight lack of movement and compensation in the upper extremities during activities of daily living. Posture, 31(3), 300–306.

27.

Murphy

M. A.

, Willén

, & Sunnerhagen

K. S.

(2013). Responsiveness of upper extremity kinematic measures and clinical improvement during the first three months after stroke. Neurorehabilitation and Neural Repair, 27(9), 844–853.

28.

Murphy

M. A.

, Willén

, & Sunnerhagen

K. S.

(2011). Kinematic variables quantifying upper-extremity performance after stroke during reaching and drinking from a glass. Neurorehabilitation and Neural Repair, 25(1), 71–80.

29.

Murphy

M. A.

, Sunnerhagen

K. S.

, Johnels

, & Willén

(2006). Three-dimensional kinematic motion analysis of a daily activity drinking from a glass: A pilot study. Journal of Neuroengineering and Rehabilitation, 3(1), 18.

30.

Osu

, Ota

, Fujiwara

, Otaka

, Kawato

, & Liu

(2011). Quantifying the quality of hand movement in stroke patients through three-dimensional curvature. Journal of Neuroengineering and Rehabilitation, 8(1), 62.

31.

Petuskey

, Bagley

, Abdala

, James

M. A.

, & Rab

(2007). Upper extremity kinematics during functional activities: Three-dimensional studies in a normal pediatric population. Posture, 25(4), 573–579.

32.

Rohrer

, Fasoli

, Krebs

H. I.

, Hughes

, Volpe

, Frontera

W. R.

, & Hogan

(2002). Movement smoothness changes during stroke recovery. The Journal of Neuroscience, 22(18), 8297–8304.

33.

Schmitz

T.J.

(2001) Traumatic spinal cord injury. In: Sullivan

, Schmitz

, (eds.) Physical Rehabilitation: Assessment andTreatment, 4th edn. F.A. Davis Company: Philadelphia, pp. 873–923.

34.

van Andel

C. J.

, Wolterbeek

, Doorenbosch

C. A.

, Veeger

D. H.

, & Harlaar

(2008). Complete D kinematics of upper extremity functional tasks. Posture, 27(1), 120–127.

35.

Van den Berg

M. E. L.

, Castellote

J. M.

, Mahillo-Fernandez

, & de Pedro-Cuesta

(2010). Incidence of spinal cord injury worldwide: A systematic review. Neuroepidemiology, 34(3), 184–192.

36.

Van Der Heide

J. C.

, Fock

J. M.

, Otten

, Stremmelaar

, & Hadders-Algra

(2005). Kinematic characteristics of reaching movements in preterm children with cerebral palsy. Pediatric Research, 57(6), 883–889.

37.

Wagner

J. M.

, Rhodes

J. A.

, & Patten

(2008). Reproducibility and minimal detectable change of three-dimensional kinematic analysis of reaching tasks in people with hemiparesis after stroke. Physical Therapy, 88(5), 652–663.

38.

Zariffa

, Kapadia

, Kramer

J. L.

, Taylor

, Alizadeh-Meghrazi

, Zivanovic

, & Steeves

J. D.

(2012). Relationship between clinical assessments of function and measurements from an upper-limb robotic rehabilitation device in cervical spinal cord injury. Neural Systems and Rehabilitation Engineering, IEEE Transactions on, 20(3), 341–350.