Machine learning, robotics and neuroplasticity

Student thesis: Doctoral ThesisDoctor of Philosophy


Stroke is a leading cause of disability worldwide. Three quarters of survivors leave the hospital with a limb impairment which makes their daily living activities difficult and reduces their quality of life. Currently, there are no drug therapies available for patients with stroke-induced limb impairments, other than intensive physical rehabilitation that require specialists and expensive rehabilitation facilities. New therapies are desperately required. Our lab and others have shown that delayed administration of neurotrophin-3 (NT-3), 24 hours following ischaemic stroke in rats, reverses the loss of motor function after 6 weeks. This is promising because NT-3 treatment has already been tested in humans in Phase I and II trials for other conditions and is safe and well tolerated. A long term goal of our research group is to understand how NT3 brings about recovery and to optimise its delivery with a view to treating humans after stroke. Further to this, preclinical assessment of therapies that might rehabilitate motor function after brain or spinal cord injury, are often done using manual behavioural test batteries, which are time consuming and labour intensive. We are leveraging robotics, computer vision, and machine learning to automate the scoring and assessment of these types of tasks. The specific aims of my PhD research are:
1. To set up in mice a photothrombotic model of large cortical stroke;
2. Assess the degree of spontaneous recovery after large cortical stroke;
3. Assess whether silencing TrkC kinase affects any spontaneous recovery;
4. Determine whether subcutaneous NT-3 improves recovery after large cortical stroke in mice;
5. Develop a device to train and assess reach-and-grasp function in mice before and after stroke, respectively;
6. Develop software to analyse mouse walking on a horizontal ladder.
Date of Award1 Jul 2020
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorLawrence Moon (Supervisor) & Dhireshan Gadiagellan (Supervisor)

Cite this