(1)
where m is the activation of the muscle function of time, c is the neutral command i-th synergy function of time, w is the constant weight of the i-th synergy referred to the m-muscle and K is the number of synergies.
The use of muscle synergies knowledge to rehabilitate postural control is still not clear. However, their role in functional movements and their importance have being already reported [10, 11]. This encourages Hyper scientific team to take them into account to promote the development of more efficient rehabilitation therapies by closely interacting with involved muscle synergies in balance control. In this way, FES can be used to develop and interact with synergies and muscle activation patterns.
3 Proposed System
In this section we describe the low-cost platform developed to perform static posturography tests and explore new therapies based of FES and muscle synergies. Balance control assessment platforms are usually not open and they are commercially available only as a posturography tool. Thus, a novel low cost and open posturography platform was developed to use it in this novel scenario. The main objective of this platform is to support the development of novel balance control rehabilitation therapies in the framework of the Spanish Hyper project.
Figure 1 shows the outline of the developed platform and a functional diagram of the different components that are further described in next sections. The platform is based on a centralized non real-time architecture, which includes several components: the Posturography Controller, the Neuroprosthetic Controller, a Wii Fit Balance Board, a Kinect camera and the TEREFES electrostimulation system [12].
Fig. 1
(a) Proposed platform architectural and (b) the functional description diagram including the different components: the posturography system (balance control assessment) and the FES system (balance control training and rehabilitation)
3.1 Wii Fit Balance Board
Wii Fit balance board is an input device included in the Wii Fit from Nintendo®;. It is a wireless device that uses Bluetooth technology to communicate with the Wii console. It is equipped with four resistive pressure sensors located in each corner of the table. In effect, it measures the displacement of the center of pressure and the weight of the user. It also gives an indication of the battery status.
Over the last years, the Wii Fit Balance Board have been used by the scientific community, specially as computer interface for disabled [13]. This device has two attractive features: it is wireless and low-cost. In our project, the Board will be used to measure the COP.
Data from Wii Fit Balance Board is accessed through a Microsoft Visual Studio C# application, using a library called WiimoteLib available at www.wiimotelib.codeplex.com. Visual C# was chosen because it is also compatible with the Kinect and its Windows Software Development Kit (SDK). Thus, the Wii Fit is connected as a HID interface device. Provided services by the Board are detected using the Service Discovery Protocol (SDP) of Bluetooth.
An important aspect to consider is the sampling frequency at which the Wii Fit sends the data to the PC, or more specifically, how often the data arrives, considering the nature of wireless transmission and the operating system behavior. To answer this question, we measured the time interval between samples using methods and public properties of the Microsoft Visual Studio C# class System.Diagnostics.Stopwatch. The program is executed in a almost dedicated HP Pavilion g6-1b70us Notebook (Intel Core i3 CPU M 370 @ 2.4 GHz, 4 GB of RAM) running 64-bit Windows 8. According to the obtained results, the average sampling frequency is 100 samples/s.
The aim of this analysis it to know how deterministic is the access to the data of the Board in terms of time. In other words, we want to know the probability that the sampling period of the Wii Fit is less than or equal to any given time. Cumulative distribution function for the Wii Fit data is depicted in Fig. 2. For example, the probability that the sampling period remains less than or equal to 0.02 s is 94.02 %.
Fig. 2
Cumulative distribution function as a function of the sampling period (s)
3.2 Microsoft Kinect
The Kinect device is a natural user interface, which allows users to interact with games without physical contact. It was developed by PrimeSense Company. The user becomes the controller itself, having to rely on movements, natural gestures and voice commands to control game elements.
Kinect is equipped with an RGB-D camera that acquires images of 640 × 480 at 30 fps. It has a visual field range from 1.2 to 3.5 m, but can be reduced by optical coupling, as Niko Zoom Lenses®;. Furthermore, its viewing angle is 57∘ horizontally, and 43∘ vertically. The vertical visual field can be expanded 27∘ with its servomotor. It is also equipped with an array of four microphones, each with a recording resolution of 16 bits sampled at 16 kHz. It also contains a stack of signal processing hardware that is able to handle all the data generated by cameras, infrared light, and microphones. By combining the output of these sensors, a program can track and recognize objects in front of it, determine the direction of the sound signals, and isolate them from the background noise.
The role of the Kinect in the platform is to enrich the visual feedback provided to the patient. Common posturography platforms are limited to provide users information about the center of pressure but the user does not know precisely his/her position and how good it is for the intended task. In this way, the Kinect provides kinematic information of full body segments, thus providing more complete information to users as well to the Neuroprosthetic Controller enabling better actuation commands.
Both, the Wii Fit as the Kinect, help to give visual feedback to the evaluated subject, and the information generated can be used by the Neuroprosthetic Controller to generate more precise and adequate stimulation patterns. A similar analysis done with the Wii Fit Board, was carried out with Kinect to evaluate the jitter effect when acquiring the frames.
The cumulative distribution function is shown in Fig. 3. For example, the probability that the sampling frequency is maintained below 35 fps is 75 %, approximately.
Fig. 3
Cumulative distribution function of the sampling period Kinect
3.3 Posturography Controller
The Posturography Controller is implemented in a personal computer running Windows 8 operating system. The developed software includes traditional posturography tools and tests like Romberg’s test, test of the stability limits and the rhythmic directional test.
The software was developed for easy use by medical personnel. It includes a database in which data of each patient is stored, allowing the physician to evaluate subject progress after several sessions. It also helps to diagnose potential diseases and program rehabilitation exercise routines for each evaluated subject. The application is also able to generate Matlab scripts containing the center of pressure points recorded during each rehabilitation session. In this way, the therapists can analyze data recorded in previous sessions.
The Posturography Controller receives all the data from the Kinect camera and the Wii Fit Balance Board. It fuses and displays the acquired data showing information like the center of pressure, the rigid body kinematic chain of the studied/analyzed subject, and information about current routines and tests. This controller and the Neuroprosthetic Controller are implemented in the same computer.
In next sections details on the parameters used to quantify the results of each test will be described [14].
Romberg’s Test. The subject is positioned on the Wii Fit Balance Board in an upright position with arms straight and close to the body trying not to move the head in neutral position facing forward, bare feet at an angle Opening of 30∘. In this position is assessed for T seconds (configurable by the doctor) their ability to maintain balance in the following conditions:
Eyes Open (REO) and Eyes Closed (REC).
Foam on Wii Fit with Eyes Open (RGA) and Eyes Closed (RGC).
The parameters evaluated in each test are:
Shift angle (degrees). The angle of the vector extending from the initial point to the subject portion to the end point of the trajectory.
Swept area (mm 2 ). It estimates the area swept by the COP by mean of an ellipse whose axes correspond to the maximum mediolateral and anteroposterior displacement.
Average speed (cm/s). It estimates the average speed, which is the ratio between the displacement and time, T, that lasts the test.
Maximum mediolateral and anteroposterior displacements (mm). These parameters represent the longest displacement in the mediolateral and anteroposterior axes during the exercise.
Figure 4 shows a screenshot during the execution of the application designed in this project. Specifically, this screenshot corresponds to a REO Test. The figure shows two visual feedbacks. The first one is the position of the center of pressure on the Wii Fit Balance Board, and the second, provided by Kinect, is the RGB image and a trace of its skeleton. The screen provides information about subject’s skeleton (skeleton blue) and a given reference (red skeleton). The reference indicates correct estimated position during the tests.
Fig. 4
Screenshot during the execution of REO Test. On the right, it is shown the center of pressure on the Wii Fit while on the left the subject with his/her skeleton (blue lines) and the given reference (red lines)
Some parameters are provided online by the application. For example, regarding the Kinect, it monitors the status of the tracking task, which can be Tracking (OK) or not skeleton (Subject not detected). Another parameter is the quality of the skeleton. This parameter indicates if the Kinect is showing the complete skeleton of the subject (Good Quality). This will help the therapist to point the camera in the correct position. Regarding the Wii Fit Balance Board, the parameters observed are subject’s weight and the coordinates of the COP.
Stability Limit Test. This test evaluates the following parameters:
LE max (mm). It is the maximum value reached by the COP in the corresponding direction (8 targets separated of 45∘ and whose radial distance from the origin is configurable).
Stability zone (cm). It is approximately the mean distance at which the patient is 90 % of time. It is calculated for each direction.
Figure 5 shows a screenshot during the execution of limit test. The displacement of the COP on the Wii Fit for the Limit Test is depicted on the right side. The red circle represents the current target to which the subject should direct his/her COP, while the green ones represent those already targeted. Traces of COPs in these directions have been deleted to not disturb the patient while reaching current target.
Fig. 5
Screenshot during the execution of the limit test. On the right side of the screen, the COP on the Wii Fit is shown in real time. On the left side is shown the patient with his/her stickman representation (blue lines)
Rhythmic and Directional Test. In this test, the patient is asked to follow the movements of a moving target (configurable frequency ) in mediolateral (ML) and anteroposterior (AT) directions. The maximum excursion limit is calculated based on the parameters of normal stability limits (previously recorded with healthy subjects).
The following parameters are evaluated for each direction.
Reaction time (s). It is defined as the time that the subject takes to bring his/her center of gravity closer than two centimeters from the reference target.
Tracking capability (%). It quantifies subject’s ability to follow the movement of the target in ML or AT directions. This parameter is calculated as the mean of error (DesiredCOP − MeasuredCOP), after the reaction time. If the error is lower than 2 cm (configurable), in other words the COP is inside the target circle, it is considered as zero for this sample.Related posts:
Merging of Humans and Machines
Pilot Study on the Feasibility of Hybrid Gait Training with Kinesis Overground Robot for Persons with Incomplete Spinal Cord Injury
Scaffolds for Neural Signal Acquisition and Analysis
Selection of Objects Combined with Autonomous Robotic Grasping
from the Past: Postprocessing of Classification Scores to Find a More Accurate and Earlier Movement Prediction
Sensor Fusion Implemented in the Posture Control of a Bipedal Robot
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
