Stroke affects approximately 150,000 people in the UK each year, leaving over 300,000 people living with moderate to severe disabilities. After stroke, many people cannot use their affected arm, and this has considerable impact on their quality of life. A major problem for stroke rehabilitation is the limited availability of physiotherapists. Therefore, home-based rehabilitation systems that do not require the presence of a therapist are needed to give the best chance of recovery through motor relearning.
Functional Electrical Stimulation (FES) of muscles is a low cost solution, which directly activates paralysed muscles through electrical stimulation via skin-surface electrodes. It has great potential as a stroke rehabilitation tool, and can even help patients with severe hand arm paralysis. In contrast to traditional physiotherapy, FES provides a means of directly tapping into the nervous system, actively producing muscle contraction and movement, exciting many of the associated neural pathways. If this is synchronised with the patient’s efforts to carry out meaningful tasks, it provides afferent inputs associated with the intention to create functional movement. This provides the most appropriate set of neural inputs to promote learning. Although recent studies have reported significant success, the problem with existing FES systems is that they require specialist skills to set up, particularly for upper limb rehabilitation, and therefore require clinical engineering involvement for each patient, negating the potential benefits mentioned above.
The challenge for the future is to enable home use without supervision by a therapist, which would require a system that could, to some extent, replace the therapist’s role of: a) monitoring the patient’s short-term progress; b) adapting the exercise regime accordingly; and c) providing the patient with real-time feedback on their performance. This is particularly challenging and the PhD project will focus on solving two connected problems:
Intelligent monitoring of patient task performance
The aim will be to use body-worn sensors during FES-supported therapy sessions to derive measures describing task performance, including movement deficiencies (poor coordination), speed of task execution, smoothness of movement, and movement variability between task repetitions (which has been shown to be a good measure of the quality of motor control). These measures will then be used for real-time biofeedback purposes, providing the patient with information on their task performance during the therapy session and, hence, to some extent replacing the therapist’s role of providing feedback.
To achieve this, powerful regression algorithms will be used to map the body-worn sensor signals onto the aforementioned variables of interest. To allow the regression algorithms to operate more successfully, be resilient to noise and variability, and adapt to different patients, fast signal pre-processing techniques will be employed prior to the regression stage.
Adaptive control of FES support so that patients are always being challenged
As the patient improves, the FES controller should adapt in real-time so that the patient must still strive to achieve the task goals. At present FES control parameters are adjusted manually, by trial and error, and it is unclear how this can be formalised so that it can be automated.
Therefore, machine leaning techniques will be applied that will use patient task performance information to make in-session patient-specific decisions on FES control parameter adjustment and, hence, to some extent replace the therapist’s role of adapting the exercise regime. Two possible approaches to solving this problem are:
i. Using a rule-based system based on the results of knowledge elicitation from a group of FES specialists.
ii. Using machine-learning methods that can automatically learn from FES specialists.
Candidates should have a first or upper second-class honours degree in an area relevant to the proposed research. This includes engineering, physics, mathematics or computer science. Candidates with other closely related first degrees should contact Prof Howard to discuss their eligibility.
For full details of student requirements and specification please visit: http://www.salford.ac.uk/ktp/industrial-case-studentships/vacancies
Informal enquiries may be made to Prof David Howard by email: [email protected]
A curriculum vitae and supporting statement, explaining your motivation and interests, should be sent to [email protected]
The studentship is fully funded and includes:
• A fee waiver
• A stipend of £15,824 p.a. for three and a half years
• All bench fees and consumable costs
• Funds specifically allocated for conference travel
Final date for applications: 6th January 2019
Interviews will be held in late January 2019
The candidate must be in a position to register by 1st April 2019