Kinematic analysis of upper extremity movement in stroke patients as a rehabilitation supporting tool

Background: Ischemic stroke often results in reduced mobility of the upper extremity and subsequent long-term disability. Evaluation of the effects of rehabilitation monitoring is often insucient, while technological progress in 3D analysis application is more objective in the assessment of motor decits. The aim of the study was to examine the use of kinematic analysis indicators for diagnosing the rehabilitation process in post-stroke patients. Methods: 20 ischemic stroke patients in the early post-stroke phase (up to three months after the stroke) took part in the study. The study tests were conducted at the beginning of the rehabilitation process and after its completion. The procedure comprised moving the index nger and reaching for four target points (closer, farther, contralaterally, ipsilaterally) placed on a table in front of the patient. The analysis of movement time and movement trajectory was carried out using the OptiTrack system. Movement time and movement smoothness (trajectory smoothness) were calculated with the use of normalized jerk score (JERK). Results: The JERK parameter changed signicantly in three movement trajectory directions: closer (p ≤ 0.01; d = 1.82), further (p ≤ 0.05; d = 1.02), and contralaterally (p ≤ 0.05; d = 0.91). Conclusions: The results conrm the usefulness of the applied measurements in diagnosing the effects of rehabilitation of patients in the early post-stroke phase. Trial registration: (not applicable)


Background
Stroke is a dangerous medical condition with serious clinical, social and economic consequences. It is the third leading cause of death and the leading cause of severe disability among people over the age of 45 [1,2]. Early and comprehensive post-stroke rehabilitation aims to reduce patients' mortality in the rst month after the stroke as well as degree of disability and life helplessness [3]. One of stroke rehabilitation goals is to minimize the risk of pathological movement patterns, excessive spasticity, or prolonged muscular hypotonia [4]. Serious post-stroke issues include limited upper limb mobility and inaccurate diagnosis of therapy effects due to lack of precise methods of therapy effectiveness evaluation [5]. There is no system for the monitoring of rehabilitation by means of various forms of electronic recording of stroke patients' progress. The assessment of motor functions in stroke patients is usually performed with standardized clinical scales. Some of the most frequently used clinical instruments for assessing upper extremity impairment and activity capacity in stroke are the Fugl-Meyer Assessment of Upper Extremity (FMA-UE) and Action Research Arm Test (ARAT) [6]. However, observer-based ordinal instruments lack the sensitivity to assess subtle, yet, potentially signi cant changes in movement performance These clinical scales tend to have a ceiling effect since they rely on scoring criteria rather than on a continuous measurement construct [7].
There have been studies of rehabilitation progress with the assistance of kinematic analysis [6,8], in particular, with robotic exoskeletons [9]. The applied devices support the motion of the shoulder girdle and the whole upper extremity. They signi cantly facilitate patients' movement and allow registering the rehabilitation progress. The study of problems related to human motor skills can contribute to the implementation of better rehabilitation at the immediate and chronic stages after stroke. The current technology offers a number of kinematic indicators, which can be e cient parameters for objective functional assessment of the upper limb in people with movement disorders [10,11]. The majority of kinematic studies on stroke patients focused on such parameters of accuracy of location in motion as close or far, low or high, or with or without clearance in front of the person. There were also studies involving different grasping tasks [12,13]. The task used in our research is similar to routine tasks of daily living, such as interacting with touchscreens or pressing buttons on different devices. Advantages of the analysis of trajectories of movement systems as a measurement tool include standardized instructions, adaptation of tasks according to patients' functioning level, and availability of quick feedback.
The aim of our study was to use kinematic analysis indicators to diagnose the rehabilitation process of stroke patients. Realizing the di culties related to the availability of equipment and analysis of results, we hope that in a few years' time kinematic analysis will become a common method of diagnosing treatment progress in stroke patients.

Equipment:
The study used the OptiTrack -Motion Capture System (NaturalPoint, Inc., Corvallis, USA) for optoelectronic motion analysis, allowing evaluating kinematic parameters of any movement. OptiTrack operates on the basis of passive markers, which during motion recording re ect the emitted IR light from seven 250e cameras. The cameras record the image of markers placed on the patient's upper body, with a maximum resolution of 832 x 832 pixels, and a frame rate of 100 frames per second. Additionally, one camera recorded video footage at 100 frames per second. This number of cameras allowed for accurate imaging of the marker's movement in space, without any loss of signal during the measurement. The data was recorded with Motive ver. 1.7.4 exported to C3D le format for processing. Markers with a diameter of 14 mm were placed symmetrically on both upper limbs. Figure 1 presents the complete set of markers in the study protocol.

Procedure:
The rst test was conducted within two days of admission to the hospital rehabilitation department. The second test was carried out after six weeks of rehabilitation in the hospital rehabilitation department. The researchers did not interfere in the rehabilitation process and did not give any guidelines for intervention.
During the test, a patient was sitting on a chair in front of a table on which four markers were placed (Fig.  2). The task performance was recorded during one test session for all participants. One trial consisted of a continuous movement toward four target points, performed at the patient's own pace. Each participant made three series, and each series was recorded separately. During the test each participant was sitting upright with the right arm abducted at a 45° angle, elbow bent at a 45° angle, and pronated forearm with the hand on the table. The left hand rested on the table outside  2. Movement smoothness (trajectory smoothness) (JERK) was quanti ed using a normalized jerk score (JERK) [15].
jerk -normalized JERK score jrk -third-order derivative of position with respect to time t -movement time a -movement amplitude "t" and "a" are the movement time and movement amplitude, respectively, applied to normalize the jerk and eliminate the in uence of movement time and distance.

Statistical Analysis
The collected data were subsequently subjected to statistical analysis using the Jamovi 1.1.9 software package. Due to the lack of normality of distributions and homogeneity of variances of the analyzed variables, nonparametric analysis tools were applied. To determine the level of signi cance of differences, the non-parametric Wilcoxon test was used to determine the dependencies between the samples. The research also used the size of Cohen's d effect. The sample size was calculated to be at least 20 participants on the assumption to detect a medium Cohen's d effect size 0.5 or larger, power of 80%, and a 5% (two-sided) signi cance level.

Results
The study results presented in Cohen's d effect size was the highest in the Closer movement direction, followed by Farther, and Contralaterally. This ranking was closely related to the degree of di culty in performing the movement in a particular direction.

Discussion
The analysis of movement trajectory (closer, farther, contralaterally, ipsilaterally) is becoming more and more relevant to the rehabilitation of post-stroke patients. This is supported by one of the hypotheses based on the theory of reorganization of cortical brain structures, which indicates the need for rehabilitation in the form of bilateral task performance involving both paretic and non-paretic cerebral hemispheres [16,17]. Quantitative motion analysis provides information on the motor compensation strategies used by stroke patients, and is therefore of considerable clinical relevance [18], as it reveals the patient's current condition and allows comparisons with healthy controls [6].
The used kinematic parameters, i.e. JERK and TIME, facilitate clinical studies of patients in terms of motion analysis, and can be useful not only for better rehabilitation planning but also for enhancing the effective application of technology-based devices [19]. One of the main advantages of this method is also the detection of individual de cits, which may remain invisible in the traditional patient assessment process. Hussain et al. [6] argue that kinematic analysis can provide valuable and speci c information about sensorimotor impairment of the upper limb following stroke that might not be captured using traditional clinical scales. Our research con rms this argument since the analysis of the trajectory of movement ipsilaterally showed no differences, which is not revealed in traditional assessment. Such information can make the patients realize the need for further rehabilitation in this area.
Trajectory analysis also makes it possible to develop a reliable reference scale for healthy people [20,21]. Otaka et al. [22] noted that arm movements in hemiparetic stroke patients were slower, less accurate, less smooth, and more segmented than in healthy controls. In the present study the imaging of movement demonstrated that toward the end of the movement its trajectory was often found to be clustered in poststroke patients, resembling a spider's web. Similar conclusions were reached by Hussain et al. [6]. Movement time and trajectory are frequently tested in fast pointing movements using motion capture systems, and are considered key variables for kinematic movement analysis of upper limb tasks in stroke [12,13]. The present study con rms their signi cance by way of performing a task of indicating a determined point by a group of post-stroke patients.
The authors of this paper also stress the possibility of using the above methodology to determine patients' individual de cits and pathological compensatory movements. The validity of determination of compensation levels and diagnosis of de cits were also discussed by other authors [8,23]. They showed that during the reaching for an object the patients displayed a reduced arm elongation and trunk axial rotation due to motor de cit. It was related to their use of compensatory strategies which included trunk forward displacement and extra head movements. The generated speci c movement compensation is closely related to the movement trajectory and the placement of objects reached for by a patient [24].
The study of movement trajectory appears to be a promising stroke assessment tool. Further research is still necessary to assess the relationship between its outcome variables and clinical measures of upper extremity impairment. This type of research proves the need to use the tools we use to support the diagnostic process of rehabilitation, however, the researchers are aware that there is still a long way to go, due to the cost and process of processing the collected information. We are also aware that the development of modern information technology will signi cantly simplify this system.

Conclusions
Movement trajectory analysis can provide more relevant information for diagnosing the rehabilitation process than procedures using traditional clinical scales.