Tongue-assisted robotics to the rescue for stroke survivors

A multidisciplinary team led by Dr Boris I Prilutsky, Dr Maysam Ghovanloo, and Dr Andrew J Butler explore the potential benefits of robot-assisted therapy over other rehabilitation methods for stroke patients with partial or complete paralysis. There is an unmet need to improve voluntary control among stroke patients by inducing self-initiated voluntary movements of the affected limbs using their tongue motion. The team has been working on improving the quality of life for these patients through robot-assisted rehabilitation.

The most common causes of paralysis include stroke (29%), spinal cord injury (SCI) (23%), multiple sclerosis (17%), cerebral palsy (7%), post-polio syndrome (5%), and traumatic brain injury (4%). These conditions impair motor neuron functions and cause people to lose voluntary control over their body, either in part or whole. Among stroke survivors, 80% have different degrees of limb paralysis, with 30% of them experiencing severe upper limb paresis (partial paralysis). This condition limits their movements, making them practically ineligible for therapeutic movement exercises to restore motor neurone functionality of the upper limb.

Rehabilitation strategies for these patients have been the topic of extensive research over the past decade, ranging from assisted physical therapy to complementary medicine. However, the most promising strategy has been robot-assisted rehabilitation therapy. To date, the net improvements in the quality of life of patients with paralysis through assistive technology have been limited, leading to estimated abandonment rates of 8 to 75% for several such technologies. However, scientists have observed a significant potential in robot-assisted therapy owing to its consistency over time and cost-effectiveness.

Innovation in robotic rehabilitation
With most assistive techniques proving to be of limited use for patients living with paralysis, scientists have sought the use of the human tongue to guide robotic rehabilitation. The tongue is superior in terms of motor control and manipulation in comparison to even the hands or the fingers. This is due to a network of neurones connecting the motor cortex with tongue motor and sensory neurons in the brainstem via neural paths. This is known as the corticobulbar connection. The hypoglossal nerve is a cranial nerve that connects the motor and sensory neurons to the tongue, lending the tongue its excellent motor control capability. The tongue muscles are also fatigue resistant and help the tongue move rapidly and freely within the oral space with little effort. Moreover, the tongue being hidden in the mouth gives the user a certain degree of privacy. In terms of assisted training, using the tongue as a control mechanism would be beneficial as it can be accessed non-invasively.

The tongue is superior in terms of motor control and manipulation in comparison to even the hands or the fingers.

Even though most people lose their ability to speak or swallow properly following a stroke, some acute and chronic stroke survivors maintain tongue function. This preservation is due to the hypoglossal nerve innervating the tongue and connecting it with the brainstem and motor cortex in both hemispheres of the brain, which escape neuromuscular injury. Survivors can thus practice tongue resistance training and improve overall tongue control. Scientists and researchers have harnessed this property of motor function conservation in several stroke survivors to overcome the limitation of existing robotic assist devices, which need a minimum level of wrist movement. Another benefit of using tongue-assisted robotic devices is that patients can use this feature even while lying down or in a posture where they feel comfortable.

The Tongue Drive System (TDS) is a wireless and wearable assistive technology that has been used as a key part of a robot-assisted rehabilitation system centred on tongue motion. TDS provides access to the voluntary motion of the tongue with the help of a small magnetic tracer temporarily attached to it. As the tongue moves, a wireless headset transmits changes in the magnetic field around the mouth that are detected by an array of magnetic sensors on both sides of the face near the cheeks to a computer or smartphone via Bluetooth. Using machine learning algorithms, the TDS headset recognises the magnetic signature of voluntary tongue motion and detects a series of user-defined tongue gestures that are translated into tongue commands. These tongue commands are then acted upon to access a PC or smartphone, drive a wheelchair, or, in this case, activate and control a rehabilitation robot system.

A team of researchers led by Dr Maysam Ghovanloo, Senior Design Engineer at Silicon Creations, and Dr Andrew J Butler, Dean of School of Health Professions at the University of Alabama at Birmingham, developed a novel tongue-operated exoskeleton system called TDS-HM. To validate its potential among stroke patients with upper limb disabilities, they coupled the TDS with the commercially available upper limb exoskeleton Hand Mentor (HM), a robotic device to control wrist and finger motion in stroke patients. The TDS tracer attached to the tongue was used to generate unique magnetic signatures for each tongue command and the headset then communicated this information to the arm exoskeleton. This helped patients perform motor-based upper limb exercises using their own volitional control.

The study involved six able-bodied individuals and three stroke patients with moderate, chronic upper limb disability. The stroke patients were given a minimal training of 15 minutes before using the device. The stroke patients had modest reductions in impairment, with functional improvement and improved quality of life, meeting the standards for the minimal, clinically relevant difference in daily-life activities, strength, and movements.

TDS – KINARM robotic exoskeleton
Together with Dr Boris I Prilutsky, Professor of Biological Sciences at Georgia Tech, the team addressed the potential benefits of tongue-controlled robot-assisted rehabilitation by developing an improved tongue-operated upper extremity robotic rehabilitation system, the TDS-KA. The TDS-KA had two units, coupling the TDS with a commercially available bimanual upper extremity exoskeleton KINARM. They chose KINARM over the HM as it could support the weight of the arm and control two degrees of freedom (providing torque to the shoulder and elbow, enabling flexion and extension) in a horizontal plane. KINARM has been used previously in neuroscience experiments to quantify motor control defects and formulate rehabilitation strategies.

The team developed an improved tongue-operated upper extremity robotic rehabilitation system, the TDS-KA.

The magnetic sensor data generated by the TDS corresponding to the tongue position were transferred to a LabVIEW based robot operator PC via Bluetooth. A support-vector machine (SVM)-based algorithm was trained to help convert this signal, with an accuracy of 93%, into discrete or proportional commands to help direct the exoskeleton robot movements via xPC target computer. At the same time, the robot operator PC controlled a virtual reality display directly and the exoskeleton robot via a xPC target computer. The xPC target computer then generated a sound-queue that translated into a specific command to be passed to the exoskeleton robot via a data acquisition board (DAQ) and connected speakers. The exoskeleton would then perform the task as commanded.

Two female stroke survivors, one with moderate and the other with severe functional impairments, were selected to assess the performance of TDS-KA by making them undertake unidirectional reaching and tracking tasks. Their Fugl-Meyer upper (FMA) extremity scores (a scoring system used to assess or quantify the degree of motor disability) were 35 and 13 out of a maximum of 66 in the beginning of the experiment. Both participants responded modestly. After six training sessions, the final FMA score of participants stood at 37 and 20 respectively. These scores indicated a clinically significant improvement in one participant (from 13 to 20). Ten healthy controls could successfully conduct their assigned tasks as well.

The team added several practical and reliability improvements, especially in the proportional control mode of the TDS through this study, which was missing in the previous studies.

Conclusion
The tongue has direct neural connections to the brain areas that not only control tongue motion but also can influence crucial functions of the limbs, a person’s memory, posture, and so on. Hence, the tongue can be used to provide neuromodulatory and reinforcement support to the brain to perform motor functions via robotic assistance.

Prilutsky, Ghovanloo, Butler and their team have successfully developed the TDS-KA system that can accurately translate tongue commands to exoskeleton arm movements, quantify the function of the arm, and perform rehabilitation training. Further iterations and larger study populations would pave the way for revolutionised robotic-assisted devices to help patients suffering from spinal cord injuries and neuromuscular diseases.