BCI-FES for Upper Limb Rehabilitation in Chronic Stroke

About one third of patients who suffer a stroke remain with a severe and permanent disability, secondary to motor sequelae and reduced mobility. Physical and occupational therapy represents the main rehabilitation alternative in patients with paretic upper limb, however, despite this, after six months of the event about two thirds of the patients will still be unable to use the affected limb in their daily life activities.

This has led to the development of new therapeutic alternatives able to enhance the effects of conventional physical therapy. The use of brain-computer interfaces represents a paradigm in neurorehabilitation of the upper limb since it can interpret the users intentions and allow them to interact with the environment by facilitating motion execution. Some effector devices that facilitate this movement execution have been tested, with functional electrical stimulation being one of the most used.

Therefore, the objective of this research is to evaluate the efficacy of an experimental therapy for motor recovery of the arm after a stroke, which includes the application of a functional electrical stimulation therapy coupled to P-300 based Brain-Computer Interface system (BCI-FES). For this purpose, the investigators will compare two groups, the first one will receive only conventional physical and occupational therapy, while the second one will receive conventional therapy together with BCI-FES therapy.

Twenty-six subjects with moderate to severe paretic upper limb and history of chronic stroke will be invited to participate in the study. These will be recruited through the patient database of the neurological rehabilitation department of the National Institute of Rehabilitation Luis Guillermo Ibarra Ibarra (INRLGII), in addition to patients recommended by the stroke rehabilitation department.

They will be randomized into two groups: the experimental group will receive P300-BCI-FES therapy together with conventional physical therapy and the control group will receive only conventional physical therapy. They will be randomly assigned to either group according to a big stick design and covariate-adaptive randomization method, considering the following variables for stratification: type of stroke (ischemic/hemorrhagic) and chronicity, baseline Fugl-Meyer Assessment Scale of Upper Extremity score, and age.

Sociodemographic variables and relevant clinical history will be collected. In addition, a series of clinical measurements will be performed through a battery of tests that will allow to know the degree of functional level of the upper limb and self-perception regarding their ability to use it in daily activities through the following scales: FMA-UE: Fugl-Meyer Assessment Scale of Upper Extremity, ARAT: Action Research Arm Test, MAS: Modified Ashworth Scale, FIM: Functional Independence Measure, MAL: Motor Activity Log. Additionally, the Patient Health Questionnaire-9 (PHQ-9) will be performed to measure and monitor the degree of depression perceived by the patients. These tests will be performed before and at the end of the intervention in both groups.

Twenty rehabilitation sessions will be performed with the P300-BCI-FES system with an approximate duration of 45 minutes, at a rate of five sessions per week for 4 weeks. It was decided to provide intensive therapy, since greater effectiveness and a higher degree of recovery have been seen in stroke patients who receive intensive rehabilitation than those who do not receive intensive rehabilitation. Together, the physician and the group of researchers from the division of medical engineering will conduct the rehabilitation sessions. The medical engineering researchers will be in charge of the technical aspects of the system, both on the FES and BCI side. On the other hand, the physician will be in charge of the patient and therapy aspects, verifying that the movements are performed as naturally as possible, and continuously assessing safety and fatigue during the sessions.

The P300 BCI-FES system, to be used for the intervention, consists mainly of two medical devices and a computer running the programs that control the operation of the medical devices and the BCI. The medical equipment consists of a biomedical signal acquisition system (g.Nautilus, g.tec medical engineering GmbH, Schiedlberg, Austria), and an electrical stimulation system (Rehastim 2, Hasomed GmbH, Magdeburg, Germany). Both devices comply with the international standard for medical devices (IEC/EN 60601).

Broadly speaking, the BCI (signal acquisition system + control software) is responsible for determining the movement selected by the patient, and for sending the corresponding command to the Neuroprosthesis (stimulator + control software), which in turn sends the stimulation sequences to the corresponding electrodes, to assist the patient's movement during the practice of upper limb functional tasks.

Patients will be asked to wear comfortable clothes, washed and dry hair, without having applied any creams or gels. A wireless acquisition system will be used to record EEG signals. EEG signals will be recorded from 13 EEG channels using the extended international system 10-20 (F3, F4, Cz, C3, C4, Pz, P3 P4, PO3, PO4, PO7, PO8, Oz, common reference in the right earlobe, ground in AFz). The programmable electrical stimulator Rehastim 2 will be used for electrical stimulation. The BCI2000 software platform will be used to perform all BCI tasks: EEG signal acquisition, ERP feature extraction, signal processing and classification, and finally prediction of the user-selected target motion. Once the target motion is selected, the BCI sends the corresponding output command to the FES control block, developed in MATLAB 2017b and Simulink, which in turn activates the FES system.

To conduct the rehabilitation sessions with the P300-BCI-FES system proposed above, a modified version of the classic matrix system P300 Donchin Speller Interface will be used, where the letters and characters will be replaced by images of hand movements and gestures including: hand opening, grasping, pinching, pronation and supination. The BCI approach will be based on the oddball paradigm, relying on conscious recognition by the user of the intensification (color change and increase in size) of a specific target containing a hand movement or gesture, within a sequence of other non-target random visual stimuli. This will elicit the P300 component in the event related potential.

A calibration session will be undertaken to obtain a first record of EEG signals that will be used to train a classification algorithm (stepwises method linear discrimination analysis, SWLDA), using the P3Classifier tool of the BCI2000 platform. The classifier coefficients obtained in this calibration session will be used during the BCI-FEST session to predict user-selected targets.

FES calibration will be performed with four pairs of transcutaneous adhesive electrodes, which are divided into two channels, one in charge of activating the flexor muscles of the wrist and fingers and the other in charge of activating the extensor muscles of the wrist and fingers. The anatomical targets of these electrodes will be the median, ulnar and radial nerves, which innervate the flexor-extensor muscles of the wrist and fingers. Channel 1 will activate the wrist flexors flexor carpi radialis and flexor carpi ulnaris; channel 2 will activate the finger flexors palmaris longus, flexor digitorum superficialis, flexor digitorum profundus; channel 3 will activate the wrist extensors extensor carpi radialis longus, extensor carpi radialis brevis, extensor carpi ulnaris; channel 4 will activate the finger extensors extensor digitorum, extensor digiti minimi, extensor indicis, extensor pollicis brevis, extensor pollicis longus.

Moreover, rehabilitation sessions will be divided into three phases. The first phase consist of action observation and target selection; in this phase the subject is asked to maintain their attention on a specific image for 20-30 seconds and count the number of times the image is intensified; they are also asked to think as if it was their own hand executing the observed movement. In phase 2 they are asked to do the same as in the first phase, but here the FES system provided passive feedback of the selected target gesture; they are solicited to observe the movement passively, without making any voluntary movements. In phase 3, they are asked to perform a hand gesture and synchronize their voluntary movements with those induced by the FES, thus allowing them to perform a functional task. During the exercises, various objects are used to recreate daily life tasks including bottles, jars, spoons, forks, toothbrushes, pencils, cotton swabs, etc.

This is repeated four times for each hand gesture and in each session the patient selectes six to eight different gestures. If the BCI-FES system correctly recognizes the P300 component, positive feedback is provided according to the selected hand gesture movement; otherwise, neutral feedback is presented and the patient is instructed to remain focused for the next exercise. The execution of hand gestures and task performance will be evaluated every two sessions, so that patients could gradually progress to more complex and difficult movements and tasks. During the intervention, the possible presence of visual fatigue, decreased concentration, muscle fatigue or pain in the upper extremity, skin irritation or redness under the electrodes, and other complications will be carefully monitored.

The benefits of rehabilitation with BCI-integrated systems can be explained by the real-time reproduction of bottom-up and top-down processes of the neural circuits involved in movement execution, thereby enabling voluntary motor control. Properties found in BCI systems that enable this include neurofeedback-associated training, reinforcement-based operant conditioning, repetitive training, and the application of the Hebbian learning principle that allows the completion of the motor intention-action loop affected by stroke.

Conventional physical will be provided over 20 sessions at a rate of 5 sessions per week for 4 weeks, to both groups. It should be noted that the control group received 1 hour of conventional therapy, while the experimental group received only 15 minutes to ensure equal dosage between the two groups. Conventional physical therapy will be focused on the proprioceptive and sensorimotor components of the affected upper extremity in muscle groups: shoulder and arm, elbow and forearm, wrist, hand and phalanges. Thus, joint mobility, range of motion, weight-bearing, sensory re-education, task-specific training and functional exercises led by an experienced professional physical therapist will be administered to both groups.

Additionally, two neuroimaging techniques will be performed including EEG and structural MRI and DTI to assess neuroplasticity after the intervention. The Imaging Division of the National Institute of Rehabilitation has a 3-T MRI scanner and the Neurophysiology Department is in charge of performing electroencephalography studies. These tools will allow us to study the morphological aspects of brain as well as the structural connectivity and functional connectivity, before and after therapy sessions, thus facilitating us to evaluate the neuroplastic changes elicited by the intervention.