Over 60 million people worldwide are blind and this number is rising as the global population ages. There are a number of major eye diseases, many of them associated with the retina and in addition many chronic diseases have manifestations in the eye. One leading cause of sight loss is Age Related Macular Degeneration, which causes severe impairment of central vision and for which there is currently no effective treatment. The complexity of the structure of the eye, the difficulty in replicating its sophisticated image pre-processing capabilities and the absence of readily available, biologically faithful models limit medical advances and treatments. At present, testing and validation of the effects of new retinal pharmaceutical interventions is primarily performed on animals, limiting the scale and scope of experiments, pharmaceutical testing and perturbations to assess what-if scenarios. Quite simply the paucity of biologically faithful in vitro or in-silico models for testing and analysis is a major stumbling block in efforts to develop effective medical interventions. Human retinal tissue is not readily available for studies. A further major stumbling block is the current and long-standing failure to fully understand retinal encoding – namely how the retina encodes image and motion information for transmission upwards along the optic nerve.
The availability of a biologically faithful in-vitro retinal model, complemented by a parallel accurate in-silico model and a new understanding of retinal image encoding of natural scenes, would accelerate the development of medical interventions to restore damaged vision in humans. Such a model would enable development of retinal prostheses, facilitate non-animal based testing of novel eye drugs and drive forward improvements in low-power, intelligent artificial vision sensors.
This PhD project is related to the in-silico component of a large integrated effort to develop and exploit a controllable and fully observable platform of real biological retinal cells for computational modelling of biological visual processing. A cross-disciplinary team of experts in nanotechnology, biology and computational modelling is driving the overall effort and establishing reliable, biologically relevant and well-characterised multi-layer 3D tissue scaffolds. These scaffold systems will have a range of applications but are particularly applicable to growth of retinal models, which will form platforms for computational modelling of brain signal processing, as well as learning and adaptation. The project will build upon our recent developments in multielectrode array recording, stimulation, analysis and spike sorting.
Scientific research questions: (a) How to progress beyond the state of the art in accurate modelling and prediction of visual signal encoding from retina ganglion cells for natural scenes; (b) developing of algorithms and approaches which explain how the brain encodes movie images in such a compact manner.
Experimental approach: A combination of algorithm development, and practical experimentation on a 3Brain Multielectrode array platform, working alongside other PhD students and research associates.
Methodology: The student will first develop a sound understanding of the state of the art in computational modelling of the retina. Following this s/he will develop artificial intelligence algorithms that seek to explain and predict how natural images and movies are encoded, initially utilising real data from recorded in a previous European framework research project but subsequently using data gathered from experiments on the multielectrode arrays of cultured retinal cells. Overall the project will involve use of artificial and computational intelligence techniques, neural network based modelling, including probabilistic neural network systems and researching computational approaches for studying synaptic modification utilising computational neuroscience learning spike sorting algorithms.
Location: Computational Neuroscience and Cognitive Robotics Group, College of Science and Technology. The CNCR lab has excellent facilities, and the group is multinational and very experienced. Nottingham is a very student oriented city in the middle of England, with two universities, with great transport links and a local airport, and economical living expenses.
The Student: This project would suit a student with good programming/modelling skills (e.g. in Matlab, Python) and a background in one of the following: computer science, physics, mathematics, computational neuroscience or a closely related discipline, together with a strong interest in multi-disciplinary modelling of brain signal processing, neural networks or cognition.
- Biologically Inspired Intensity and Depth Image Edge Extraction, IEEE Trans Neural Network and Learning Systems 10.1109/TNNLS.2018.2797994
- Computational modelling of Salamander Retinal Ganglion Cells using Machine Learning Approaches, https://doi.org/10.1016/j.neucom.2018.10.004
- Bioinspired approach to modeling retinal ganglion cells using system identification techniques. IEEE Trans Neural Network and Learning Systems, 10.1109/TNNLS.2017.2690139
Entrants must have a first/undergraduate Honours degree, with an Upper Second Class or a First Class grade, in computer science, physics, mathematics, computational neuroscience or a closely related discipline. Applicants with a Lower Second Class grade at first degree must also have a postgraduate Masters Degree at Merit or Commendation.
Fees and funding
This is a self funded studentship opportunity.
Guidance and support
Further guidance and support on how to apply can be found on this page.