Novel methods for the restoration of upper limb and hand motor function Silvestro Micera Scuola Superiore Sant Anna, Pisa (I) Swiss Federal Institute of Technology, Zurich (CH)
Outline of the presentation New approaches for neurorehabilitation The robotic gym New protocols Synergies among different technological solutions Error-enhancing protocol for neurorehabilitation A wearable system for FES Conclusions and future works
Outline of the presentation New approaches for neurorehabilitation The robotic gym New protocols Synergies among different technological solutions Error-enhancing protocol for neurorehabilitation A wearable system for FES Conclusions and future works
Classification of machines for neurorehabilitation Exoskeleton-like machines: Application to patients with severe disabilities (when single joint control is required, absence/very few motor sinergies) Class I Mechanical/Hydraulic/Pneumatic actuation High power, very precise Heavy, non-portable Class II Wearable, portable systems Low power, limited precision Operational-type machines: Application to patients with moderate disabilities (when the patients feature a sufficient level of natural motor sinergies) Class I Low mechanical inertia/friction High back-driveability Fine tuning of viscoelastic properties for force fields generation and measurement of the impedance of the human arm Class II Simple mechanical structure, no back-driveability Active compensation of inertia/friction Micera et al., 2005
Exoskeleton-like machines The machine is designed so that the trajectories of its end-effector AND of ALL its joints are equal to that of the natural limb in the operational space AND in the joint space
Operational Type Machines The contact between the patient and the machine is only at the end effector, through a purposive mechanical interface (e.g. pedal or handle) The machine is designed so that the trajectory of its end-effector is equal to that of the natural effector (hand/foot) in the operational space The patient is expected to exploit her/his own synergies at joint level to follow a trajectory in the operational space The MIT-MANUS system (Inmotion Ltd.)
Class I and II Operational Machines Among the operational machines, two different classes of devices can be identified Class I systems (Volpe et al., 1999) characterized by a low mechanical inertia/friction, a high back-driveability, fine tuning of viscoelastic properties for force fields generation and measurement of the impedance of the human arm, and high cost Class II (Reinkensmeyer et al., 2002) systems characterized by a simple mechanical structure, no back-driveability, (in some cases) an active compensation of inertia/friction and a low cost Even if Class II operational machines present some limits, they are very interesting because the low- cost and the simplicity of functioning can make them more acceptable in clinical practice and even for telerehabilitation The potentials of these simple machines in terms of functional recovery are to be analysed
The robotic-gym for neurorehabilitation Moderately disabled subjects Severly disabled subjects Exoskeleton (Partial) Motor recovery Operational Class I robots At the hospital Operational Class II robots for telerehabilitation Micera et al., 2008
Clinical Validation of the MEMOS I system Starting position P2(X2,Y2) Final position P1(X1,Y1) Clinical trials (2003 present) at Fondazione Maugeri, Veruno (Italy) Drs. Pisano and Colombo Micera et al., 2005; Colombo et al., 2005; Colombo et al., 2008
Clinical validation PRE-REHABILITATION REHABILITATION POST-REHABILITATION An example of trajectory for subject S1 before and after the treatment The activity carried out by the robot is underlined
Clinical validation An example of tracking of the squared trajectory for one subject
Clinical assessment scales The robot-assisted therapy was accepted and well tolerated by all the patients included in the study; no adverse events were registered
Robot-derived assessment parameters Example of the time course of the motor recovery components assessed by the evaluation metric One can note that the AMI increases up to half-way through treatment when the patient is able to complete the motor task The mean speed VM is constantly increasing, indicating continuous improvement of the patient's performance throughout the treatment. The mean distance (MD) and the normalized path length (npl) decrease, thus showing an improvement in both accuracy and efficiency of the movement The npeaks show the continuous improvement of movement smoothness The nfcp shows an improvement of force control, indicating a positive change in the movement dynamics
MEMOS II
Tele-rehabilitation using the MEMOS Delay Modification of the protocol Warnings Assessment parameters Internet Assessment parameters Delay Modification of the protocol Warnings Therapist at the hospital (checking the safety of the experiments, the values of the assessment parameters, the need for a change of the protocols, etc.) Patient at home Modified from Carigan and Krebs, JRRD, 2006
New protocols? Standard robotic aid therapy Similarly to therapist handover-hand assistance during conventional therapy Highly task oriented practice environments Active resistive exercises 2 Virtual reality 3 Different training approaches Different biofeedback 1 EMG-based control of pointing movements Recording of EEG signals For highly functional patients, because of the ceiling effect in the learning process, no improvements could be possible Customized therapy depending on patient injury level The robot assisted motion when patient could not complete the task 1 DiPietro et. al. 2005, 2 Krebs et. al. 2003, 3 Merians et al 2002, 4 Patton et al 2005
Synergies among different technologies Robotics and FES are two complementary rehabilitation technologies which can be used together To restore different motor functions (e.g., hip- knee using robotics, ankle using FES) To restore the same motor functions in a customized way for the different patients
Outline of the presentation New approaches for neurorehabilitation The robotic gym New protocols Synergies among different technological solutions Error-enhancing protocol for neurorehabilitation A wearable system for FES Conclusions and future works
Exploiting the potentials of motor learning in neurorehabilitation There is a general consent on the theory stating that, when human subjects are asked to move in new dynamic environments, an Internal Model of the external world is generated and/or updated by the CNS to achieve the desired trajectory of the arm Motor leaning is fundamental in neurological rehabilitation Recent studies (Patton et al., 2006) involved the use of adaptive training techniques with hemiparetic stroke patients, and concluded that an amplification approach provides a new pathway for augmenting motor learning in individuals with brain injuries
Exploiting the potentials of motor learning in neurorehabilitation This kind of approach may induce the CNS to attempt a new motor strategy The change in reflex tone leads to a better movement, so that if a spastic muscle pulls the limb to the side and the robot pushes the arm to increase the error, the spastic muscle would be shortened An adaptive training could lead the CNS to promote learning by making errors Augmenting errors may be also correlated with motivation and attention, and it can increase the signal to noise ratio for sensory feedback and self evaluation
Divergent force fields in able-bodied subjects Divergent force field Humans can learn to make accurate movements by controlling magnitude, shape, and orientation of the endpoint impedance 1. Burdet et al., Nature, 2001
A NOVEL ERROR ENHANCING ROBOTIC-AID THERAPY ERROR ENHANCING THERAPIES Comparison between the outcomes of the classic active assistive robotic therapy and a new error enhancing therapy We are investigating whether: 1) hemiparetic subjects are able to adapt to unstable dynamics 2) the use of this new protocol could enhance motor recovery 3) it provides a better outcome when compared with the assisted-as-needed rehabilitation therapy Burdet et al, 2001
A NOVEL ERROR ENHANCING ROBOTIC-AID THERAPY TRAINING PROTOCOL 6 weeks of therapy: 2 weeks (10 days, 1 hour session each) - first therapy cycle 2 weeks break 2 weeks (10 days, 1 hour session each) second therapy cycle 9 turns of the game, being trained with active assistive or DF field, with 1 turn in Null field (NF) conditions DIVERGENT FIELD/ ACTIVE ASSISTIVE (GROUP 1) ACTIVE ASSISTIVE / DIVERGENT FIELD (GROUP 2) Three different magnitudes of the divergent field (high, medium, and low) During each day of DF therapy, the hand was deviated initially using a low intensity field, then a high one and finally a middle one.
A NOVEL ERROR ENHANCING ROBOTIC-AID THERAPY ASSESSMENT OF RECOVERY COMPONENTS Motor Status Score (MSS) CLINICAL SCALES Evaluated before and after each therapy cycle Modified Ashworth Scale (MAS) Range of Motion (ROM) Chedoke Master Stroke Assessment (CM) NPeaks - Number of peaks in the speed profile REACHING INDEXES Used in NF conditions Smoothness The Teulings parameter npl - Path Length Parameter MVD - Movement direction variability S * duration length 5 2 = J dt 2 ABSOLUTE HAND PATH ERROR (AHE) Used in DF conditions
A NOVEL ERROR ENHANCING ROBOTIC-AID THERAPY RESULTS GROUP 1 gradually become proficient at producing straighter trajectories: they learned how to contrast the field GROUP 2 presented a more discontinuous trend: lower decay rate and a not significant correlation coefficient of the regression
A NOVEL ERROR ENHANCING ROBOTIC-AID THERAPY RESULTS Significant reduction in impairment of the hemiparetic limbs, as shown by the evolution of the MSS and MAS throughout the therapy The lower the impairment, the bigger the improvement
A NOVEL ERROR ENHANCING ROBOTIC-AID THERAPY RESULTS Variation of the number of peaks depending on both the therapy and the patient severity level 80 70 60 50 40 30 20 10 0 I II I II I II I II I II I II 6 4 3 6 4 3 Chedoke stages Active assistive CF Chedoke stages DF DF The application They of DF benefit seems more to if trained be more effective in the in patient active assistive with than lower upper limb in impairment DF score
A NOVEL ERROR ENHANCING ROBOTIC-AID THERAPY DISCUSSIONS Post stroke patients were able to contrast the perturbation They could reach the target end perform the exercise but depending on the level of impairment the ability in contrast the perturbation changed Robotic-aid therapy led to significant improvements in both cases DF seems to have no negative effects Different effects have been observed between different injury level patients and therapies Application of the DF at first could determine a stabilization of the posture Divergent/active assistive seems to be better to active assistive/divergent therapy Does fatigue affect the outcome of the results? A 3 month follow up will point better the differences between the two approaches DF/active assistive more effective on mild moderate patients; active assistive/df more on severe patients Customization of therapy For severe pathological subjects interacting with unstable dynamics and motor variability can led to undesired outcomes
Outline of the presentation New approaches for neurorehabilitation The robotic gym New protocols Synergies among different technological solutions Error-enhancing protocol for neurorehabilitation A wearable system for FES Conclusions and future works
Main Goal To develop a Neuroprosthesis (NP) garment based on novel textile electrode technology to restore hand function Deliverables NP garment with integrated [multi-channel] electrode pads, sensing and stimulator Methods for garment integrated sensing & NP control Sensing: EMG, Flexion etc Processing: Muscle activity & fatigue, User interaction, control cmds Advanced control algorithms Targeted Population Stroke: Largest group, >10,000 / year (USA) Spinal Cord Injury (SCI): small group but could benefit
Requirements: Functionality Grasp functions to assist Activities of Daily Living (ADL) Cylinder (volar) grasp [Light 2002; Sollerman 1995] Lateral (key) grasp Opposition (pulp pinch) grasp Stabilise wrist extension (>20±3 ) Easy to (re) configure grasp function for different patient / user Electrode re-configuration during early treatment Multiple electrodes / channels Adjustable control of timing for grasp function. Adaptable configurations Should easily integrate with patient intention Easily select / start / stop different grasps Goal orientated MMI Enable home-based patient treatment programs Provide muscle strengthening protocols training Help reduce spasticity Muscle strength Spasticity therapy
Introduction to TES Natural muscle movement Action potentials (AP s) from motor cortex AP s propagate along spinal cord Produce contraction of muscle fibres Transcutaneous (surface) Electrical Stimulation (TES) Activates motor-neurons with electrical pulses Delivered between pairs of electrodes Electrode placement is critical for selective muscle activation
Introduction to TES Self Adhesive Transcutaneous Electrodes Flexible conductive material Stainless steel mesh, carbonized rubber Skin interface of conductive hydrogel Gelatinous adhesive electrolyte <1mm Anode (+) Cathode (-) Sensory receptor activation Discomfort during TES Multiple muscle activation Lack of selectivity Accurate cathode placement
Objectives Improve Selectivity Simplify Application Integrate Cabling Improve Comfort
Overview Improve Selectivity Selective muscle activation using TES? Effect of cathode and anode positions? Can we achieve selective finger activation? Simplify Application Integrate Cabling Embroidered Electrode Technology Can embroidered electrodes be used for TES?
Measuring Finger Selectivity [Lawrence,2007] No standard device for assessing isometric finger forces AND wrist torques Developed Grasp Force and Wrist Torque Assessment System 5 load cells to record isometric finger forces (A) 6-dof load cell to record isometric wrist torques (F) 3D measurement system for anatomical landmarks (C) Integrated with Virtual Electrode Environment Includes 64 element multiplexer for arrays Embedded PC running xpc real-time OS Data logging & array control
Selective Activation using TES? Can middle & ring finger be selectively activated? Small probe (Ø=3mm), 11 11 grid, 5mm spacing What is influence of pulse width (PW) on selectivity? 200µs Selective activation is possible Coupling observed 500µs Higher PW increases coupling, reduces comfort Use shorter pulses
Effect of cathode position Place arrays above extrinsic flexors and extensors 30 elements (~12 12mm), hydrogel interface Dynamically switch cathode across array surface; Anode elements remain as far away as possible Extrinsic flexor maps Extrinsic extensor maps Selective middle and ring finger flexion Coupled wrist flexion Coupled middle and ring finger extension Coupled wrist extension Selective finger activation; but coupled with wrist torques
Selective finger flexion [Lawrence, 2008] Co-activate extensors to compensate for wrist torques Functional grasp Selective finger activation Selective finger flexion requires co-activation of extensors
Overview Improve Selectivity Use shorter pulse widths ~200µs Anodecan be placed arbitrarily Selective finger flexion requires co-activation of extensors Simplify Application Integrate Cabling Embroidered Electrode Technology Can embroidered electrodes be used for TES?
Embroidered Textile Electrodes KTI Smart Electrodes 7735.1 DCS-LS & 9005.1 PFLS-LS Conductive yarns Embroidered electrodes Embroidered arrays & cables Direct stimulation was painful; need hydrogel layer
Prototype NP Design Optimised electrode positions for Cylindrical, Lateral, Opposition grasps Size, shape, orientation of activation regions adapted using hydrogel pads Embroidered, machine washable garment with integrated electrodes + cables Anode EMG Ref Extensors Thumb adductors Thumb flexor Index Flexors Clinical testing starts soon
Main Contributions: Results Transcutaneous Electrode Technology for Neuroprostheses Improve Selectivity Simplify Application Integrate Cabling Improve Comfort Multiple Electrode Arrays Selective finger activation Co-activate flexors & extensors Embroidered Electrodes Suitable for use in arrays; enables integrated wiring; requires use of hydrogel Electrode Comfort Comfort related to contact area, not resistivity Dynamic anode placement Simplifies array design; Continuous hydrogel layer Complex multiplexer Element Area >1cm 2 Improves comfort; Enables selective activation Reduces resolution
Clinical trials The wearable devices will go into clinical trials in the next months: Balgrist Hospital and ZAR, Zurich University of Southampton REL, University of Toronto Particular attention will be devoted to the combination between robotics and FES for upper and lower limb function restoration
Outline of the presentation New approaches for neurorehabilitation The robotic gym New protocols Synergies among different technological solutions Error-enhancing protocol for neurorehabilitation A wearable system for FES Conclusions and future works
Conclusions Error-enhancing protocols could be used with interesting results (especially in people with mild impairments) More extensive clinical trials are necessary to confirm these results It would be important to define customized strategies to provide errorenhancing and assisted-as-needed trials
Conclusions Individual limitations of the robotic and FES therapies can be eliminated by combining the two modalities Immediate advantages include promotion of normal muscle activation, the possibility for practice of normal patterns earlier during rehabilitation, reduced requirements on physical therapist support, and ankle/hand activation Wearable systems could address some of the limits of current FES systems
The robotic-gym for neurorehabilitation Moderately disabled subjects Severly disabled subjects Exoskeleton (Partial) Motor recovery Operational Class I robots At the hospital Micera et al., 2008 MEMOS II will be tested in a network of clinical ( GIOMI ) centers Operational Class II robots for telerehabilitation
Next future?? Brain injury / Neurological impairment Off-line On-line brain imaging (assessment) (Sensory-) Motor impairment Recovery of motor function (motor outcome) Recover of brain functions (through motor learning and plasticity) Limb motor rehabilitation Neurorehabilitation Cognitive rehabilitation
Epidural neuroproshesis for locomotion Project coordinated by Dr. Courtine (UZH)
Thanks for the attention!! Email: micera@sssup.it