WO2005011494A1

WO2005011494A1 - Method and biomechanical device for canceling pathological tremors

Info

Publication number: WO2005011494A1
Application number: PCT/ES2004/070055
Authority: WO
Inventors: José Luis PONS ROVIRA; Eduardo Rocon De Lima; Leandro Bueno; Andres F Ruiz Olaya; Leopoldo CALDERÓN ESTÉVEZ; Ramón CERES RUIZ; Antonio R JIMÉNEZ RUIZ
Original assignee: Consejo Superior De Investigaciones Científicas
Priority date: 2003-07-25
Filing date: 2004-07-21
Publication date: 2005-02-10
Also published as: ES2222819B1; ES2222819A1

Abstract

The invention relates to a method enabling selective biomechanical cancellation of pathological tremor. Canceling focuses on the characteristic frequencies of the tremor and has no negative effect upon the voluntary movement of the patient. It can be used in different types of tremors and situations by automatically adapting to the specific characteristics of the tremor signal. The whole system takes the shape of an upper limb orthosis and, therefore, it can be used in a series of day-to-day activities. The characteristics of commercial products are more restricted: they are limited to concrete activities (feeding), they apply non-selective biomechanical action and cannot therefore adapt to the concrete characteristics of the tremor or patient, they are desktop devices and are thus not portable.

Description

TITLE

METHOD AND BIOMECHANICAL DEVICE FOR CANCELLATION OF PATHOLOGICAL FEAR

TECHNICAL SECTOR

The present invention relates to biomechanical or biomedical devices for the cancellation, reduction, or control of involuntary pathological tremor. As such, they are applicable in the prosthetic and orthotic rehabilitation industry.

STATE OF THE ART

Pathological tremor is an effect of some neurological diseases that results in the disability of the affected person. It is characterized by rhythmic and involuntary contractions of the patient's muscles. As a consequence the affected person is subjected to involuntary oscillations of his upper limbs and possibly other parts of the body.

The frequency of these oscillations varies depending on whether the type of tremor is resting (characteristic of Parkinson's disease), postural (characteristic of patients with essential tremor) or kinetic (characteristic of patients with traumatic brain damage). The commonly accepted range ranges from 2Hz to 12Hz.

The classic approach to treating tremor is medication and electrical stimulation of some areas of the brain. However, there are other devices that act by applying a resistive or dissipative mechanical load on the affected organ. Such is the case of the device marketed under the name "Neater Eater". The state of the art is such that a dissipative mechanical load is applied to these devices independently of the type of tremor affecting the patient. This resistive load is also independent of the type of movement on which it is applied, that is, whether it is voluntary, trembling or a combination of both. How Consequently the patient notices a resistive load also in the face of his intentional movement.

In the state of the art there are some patents whose object is to reduce the effect of tremor on specific patient activities. Such is the case of patent US6561993 which focuses on the filtering of the shaky signal in input peripheral devices for computers. The object in this case is not to alter the tremor pattern itself, but to prevent the trembling signal from being reflected at the output of the computer's peripheral. Other patents, such as US5293879 or US 5964720, focus on procedures for detecting tremor as a consequence of neurological diseases and procedures for generic physiological monitoring and in particular for monitoring tremor.

The present patent relates to a selective tremor cancellation method. The method is structured around a repetitive control loop. In the state of the art there are several embodiments and patents that employ or focus on this type of control loop. Specifically, patent US5740090 focuses on the protection of a type of filter tuned with a main frequency component, "notch filters", of order equal to or greater than two, and of repetitive control and learning strategies developed from said filter. This patent also cites other previous patents, such as US4821168, which focuses on robustness aspects of the aforementioned control loops.

DESCRIPTION OF THE INVENTION

BRIEF DESCRIPTION OF THE INVENTION Pathological tremor is characterized by rhythmic oscillations of some member of the human body. These rhythmic oscillations are superimposed on the patient's voluntary movement. The amplitude and frequency characteristics of the trembling oscillations depend on the type of tremor, be it resting, postural or kinetic. In addition, for the same patient, these characteristics may be altered depending on the activity or conditions that he develops or is subjected to. A clinically established procedure to reduce the effects of tremor is the application of resistive mechanical loads, whether these are dissipative or inertial. The ideal solution is to apply loads of this type only to shaky movement. However, all the embodiments described in the state of the art apply the biomechanical load to the combination of the voluntary and trembling movement. This results in resistance to the patient's intentional movement, and as a consequence a lack of comfort.

Additionally, there are no methods in the state of the art that allow the specific application of loads depending on the specific characteristics of the type of tremor (variable frequencies and amplitudes) or the particular characteristics at a given moment of the tremor in a patient, which they will generally be different from those of any other instant.

The present patent aims to protect a method that allows the selective cancellation, reduction or suppression of tremor, that is, without affecting the voluntary movement of the patient. Furthermore, the proposed method allows this cancellation action to be adapted automatically to any type of tremor or to any tremor condition of a patient.

The proposed method basically consists of two stages: 1. The first stage is a strategy to identify the characteristics of the tremor based on information collected by a variety of biomechanical (position, speed, acceleration, or force) or physiological (signal) sensors. electromyographic). The characteristics obtained by this strategy are at least the frequency, phase and amplitude of the trembling signal, differentiating it from the voluntary signal, using any digital algorithm or digital or analog electronic circuit designed for this purpose. 2. The second stage is made up of a repetitive control loop, which tuned with the tremor information obtained in the previous phase to the specific characteristics of the trembling signal, selectively cancels the trembling part of the movement. For this, a multiplicity of actuators allow the cancellation action of the control loop to be applied, through the application of internal forces (solution orthotic) or by applying external forces (tabletop solution, for wheelchairs ...).

Using this method, only the trembling part of the patient's movement is selected for subsequent cancellation. Since the first stage is of a general nature, the identification and monitoring of the trembling signal allows its application to any type of tremor or to any condition or activity of the user.

DETAILED DESCRIPTION OF THE INVENTION Pathological tremor presents a rhythmic oscillation of the different segments of the upper limb of patients with some diseases of neurological origin. This oscillation is characterized by frequencies in the range of 2 to 4 Hz for the case of resting tremor characteristic of Parkinson's disease. Postural tremor is however characterized by a frequency range of 5 to 8 Hz and is characteristic of essential tremor disease. For its part, kinetic tremor can occur at frequencies between 4 and 12 Hz and is characteristic of patients with traumatic brain damage.

Given a patient affected by any of the tremors described above, its specific oscillation characteristics may vary depending on the activity or condition it supports. These oscillations will generally be superimposed on the patient's intentional movement.

Voluntary movement is generally characterized by significantly lower frequencies than tremor. Furthermore, the latter is an essentially harmonic (rhythmic) movement. Taking advantage of these two characteristics, it is possible to determine strategies that allow differentiating both types of movement. Specifically, one possible strategy is: a) filtering of the motion signal with low or high pass filters designed with cutoff frequencies in the frequency range between voluntary and shaky motion. Filters thus designed can be implemented both digitally and analogically and can be of any order, that is, with rejection slopes at the highest or lowest prohibited frequencies.

Taking advantage of the essentially harmonic characteristic of the tremor, another procedure for the identification and monitoring of the tremor signal is based on: b) its modeling by means of a series development of a main harmonic of frequency f ₀ and secondary M, as reflected in the equation (1).

- sm rf _Q k + wr + Mk eos rf ₀ k] r = l

The error, β _k , can be established between the model of equation (1) and the signal actually obtained through a variety of sensors at time k, s _k , according to equation (2). ^e k = ^S k ^~ l ^w rk ^{• sin} KÍθ ^k + ^Wr + ^{Mk C0S} * fθ ^k (2)

This error can be used to propose a least squares optimization process for the model parameters of equation (1). The described process allows, through the proper selection of the convergence parameters of the optimization process, the obtaining of a recursive method for the instantaneous estimation of the frequency, phase and amplitude characteristics of the trembling signal.

The described strategy allows obtaining an instantaneous estimate of the characteristics of the tremor, differentiating these from the voluntary movement. Although two alternatives have been described for the implementation of this strategy, it is the strategy itself that allows the implementation of the method protected in this patent and as a consequence combinations of both or third digitally developed methods can be used in the domains of time or frequency or analog.

Repetitive control is a variant of learning control in which prior information on the type of disturbance to which a child is subjected is used. system to improve rejection of them. Specifically, it is applicable when the disturbances are periodic.

In the case of pathological tremor, the system to be controlled is the upper limb of the patient and the disturbance is the oscillation caused by the tremor and which, as indicated above, is rhythmic (harmonic). Therefore, selective tremor cancellation or rejection can be approximated by repetitive control loops.

In repetitive control, a linear combination of one or more previous error values between the reference and the measured signal is used to construct the input signal to the system controller, see Figure 1. The result is that the system transfer function control presents a very large attenuation at the fundamental frequency of the disturbance and in its corresponding harmonics as indicated in figure 2. The number of previous error values and the coefficients of the linear combination described above determine the width of the attenuation zone and I reject.

The strategy described in the previous paragraphs allows, once the main frequency of the tremor is known, to perform a selective cancellation of the disturbance represented by the tremor to the voluntary movement. If, as described above, the identification strategy allows instantaneous identification of the tremor characteristics from the information from the sensors, the proposed method automatically adapts to the type of tremor and the patient's conditions.

The implementation of the control method can be carried out both digitally and analogically without affecting the basis of the protected method. The control action described above results in the application of a series of biomechanical stresses to the trembling limb. These efforts can be internal, that is, applied to the limb taking another segment of the human body as a reference, leading to orthotic solutions. Efforts can also be external, that is, taking as reference any external element (table, wheelchair ...), leading to non-ambulatory applications. The described method is shown schematically in the diagram of figure 3. Therefore, it takes a stage of obtaining sensory information, a strategy for identifying and monitoring the characteristics of the tremor from sensory information, a repetitive control loop for the rejection of the disturbance due to the tremor and a multiplicity of actuators for the physical application of the control action.

EXAMPLE OF IMPLEMENTATION OF THE INVENTION

The development of an orthotic solution to cancel tremor at the elbow is presented as an example of embodiment of the invention. The orthosis is formed by a rigid structure that allows supporting a set of sensors and actuators to implement blocks 1 and 4 of the cancellation method described schematically in Figure 3.

In the example, two gyroscopes located on both sides of the joint are used as sensors that provide information on the absolute angular velocity of the arm and forearm in the plane of movement of the joint. The measurement frequency range is between 0 and 50 Hz, thus covering the entire frequency range of possible tremors. An ultrasonic motor is also included to generate the tremor cancellation torque at the joint of interest. The motor torque is 1 mN, enough to counteract the average torque of the tremor at this level.

The set is controlled by a personal computer that incorporates acquisition cards of sufficient capture speed and resolution, in this case 200 Khz. and 12 bits respectively. A tremor identification and monitoring strategy is implemented according to the IEEE-STD-1057 standard to develop block 2 of figure 3. This strategy assumes that the tremor follows a harmonic pattern and develops a model of it according to the equation (1). The recursive least squares optimization of the error between this model and the actual measurement of the tremor allows instantaneous identification of the frequency, phase and amplitude of the tremor. The selected sample rate for the identification loop is 1 Khz. Since the selected strategy gives the estimate with a single sample delay, the estimate delay is reduced to 1ms, negligible compared to the intrinsic frequency of the tremor that is between 2 and 12 Hz The estimated frequency of the tremor is used to tune the repetitive control loop of block 3 of figure 3. A vector is used for this, in which the error between the reference and the signal in the previous period is stored. In this case, a linear combination is used with a single previous value, the one corresponding to a one-period delay of the identified tremor signal. This linear combination is the input to a classic integral proportional controller.

DETAILED EXPLANATION OF THE DRAWINGS

Figure 1 The figure schematically represents a repetitive control loop. Block 1 represents the plant to be controlled, which in this case is made up of the whole of the member and the device used for applying forces. Block 2 represents the controller that, based on the error signal e _r , generates the control action. Block 3 develops a linear combination of M previous values of error e. In this scheme the error e is determined from the difference between the reference value r and the measured signal s.

Figure 2

The figure represents the control action represented by a repetitive loop. The tremor center frequency, f ₀ , is used to tune this repetitive loop resulting in a large attenuation or rejection centered on this frequency and the corresponding harmonics.

The selection of the central rejection frequency, fo, implemented by the repetitive control loop according to the instantaneous characteristics of the tremor allows the biomechanical cancellation, suppression or reduction of the tremor independently of its particular characteristics. Figure 3

The figure represents the set of steps necessary for the implementation of the method described in this patent. Block 1 is made up of a variety of sensors that collect kinematic or kinetic information about the tremor in a given limb (position, speed, acceleration, force or mechanical torque) and / or physiological information about it (electromyographic signal). Block 2 represents the strategy for identifying and monitoring the characteristics of the tremor, using as input the sensory information provided by block 1 and generating an estimate of the frequency, phase and amplitude of the trembling part of the movement of the affected limb. Block 3 represents the application of a repetitive control loop tuned with the information provided by block 2 to implement a rejection zone of frequencies adjusted to the characteristics of the tremor. The control action represented by block 3 is the input of block 4 that represents the biomechanical application of the action of cancellation of the tremor by means of a multiplicity of actuators arranged in a system of internal (orthotic) and / or external efforts (desktop devices). , wheelchair...)

Claims

1.- Method of cancellation, suppression or reduction of pathological tremor in any of its variants characterized by: a) Strategy for identification, characterization and monitoring of the tremor in real time in terms of its main frequency, phase and associated amplitude. b) Repetitive control loop tuned according to the characteristics provided by the system a) for the generation of the control and cancellation action of the tremor. c) A multiplicity of sensors and actuators that allow obtaining the necessary information about the tremor in accordance with strategy a) and that allow acting on the affected organ in accordance with the control action b).

2. Tremor identification strategy based on the method described in claim 1 specifically in point 1a) which is carried out by means of an algorithm in the frequency or time domain and which is implemented in a controller program.

3. Method described in point 1 in which the strategy of identification of the tremor of the point based on claims 1 and 2 is performed electronically by means of an electronic circuit for that purpose.

4. Control loop based on item 1b of claim 1 which includes one or more harmonics of the fundamental frequency of the tremor and which is implemented digitally or analogically. The repetitive control can be implemented by using one or more previous values of the controller error.

5. Process of identification of the characteristics of the tremor based on the method described in point 1 a) of claim 1 and which is carried out in real time, providing instantaneous information or with a delay that does not affect the stability of the loop described in point 1 b) of claim 1 and allows its application to any variety of pathological resting, postural or kinetic tremor, and under any condition of patient activity.

6. Method described in point 1 and its Development of both orthotic devices, that is portable, as well as non-ambulatory desktop devices, for wheelchairs or any additional device that in practice employ patients with pathological tremor based on the methods described in claims 1, 2,3,4 and 5.