MRes students work on their research project throughout the year.  Projects available for 2021-22 will be shown here soon.   You can apply for one of the projects listed here, OR contact a supervisor to develop a different project.

Please for full details on how to make your application.

Projects available for 2021-22

Previous MRes Neurotechnology Projects (for information only)

Some of the MRes projects from previous years are shown here to give an idea of the topics covered by our students.

Previous projects

Auditory blast biomarkers

Supervisors: Andrei Kozlov (Bioengineering), Tobias Reichenbach (Bioengineering)

The project will focus on using EEG data recorded from healthy and blast-TBI rats to discover electrical signatures of auditory cortex damage. Furthermore, extracellular multielectrode array recordings will be used simultaneously with EEG, and the student will attempt to develop an algorithm to predict multiunit activity from EEG. The ultimate goal of this work is to develop a set of electrophysiological markers for auditory processing disorder, a common debilitating condition in people exposed to blast (such as war veterans). The student will work in collaboration with other members of the Kozlov lab working in this area. Dr Reichenbach and I have an on-going collaboration on this topic, and this project will be a useful addition to our development of the data analysis pipeline. As for the interdisciplinary aspect of the project, Dr Kozlov’s lab will bring expertise in auditory cortex neuroscience in animal models, whereas Dr Reichenbach will contribute invaluable knowledge of EEG analysis methods used in humans. Our common goal is to bridge the gap between the animal (invasive) and human (non-invasive) studies of auditory processing disorders.

Classification of EEG responses to multi-dimensional transcranial electrical stimulation

Supervisors: Gregory Scott (Brain Sciences), Ines Violante (second supervisor tbc)

Background
A major shortcoming of medical practice is the lack of an objective measure of conscious level. Impairment of consciousness is common, e.g. following brain injury and seizures. Brain injury can also interfere with sensory processing and volitional responses, which confounds the behavioural assessment of conscious level. Any objective measure of consciousness would therefore ideally bypass sensory processing. One such example is the perturbational complexity index (PCI) [1], which quantifies the average complexity of brain activity to multiple pulses of transcranial magnetic stimulation to provide a uni-dimensional measure of conscious level.

Proposal
We are interested in developing a novel measure of brain state derived from the set of brain responses evoked by a multi-dimensional (i.e. potentially varying in electrode space and time) ‘program’ of transcranial electrical stimulation (TES). 

We have recently acquired the GTEN neuromodulation system, state-of-the-art hardware which allows TES and EEG to be recorded through the same high-density (256 electrode) cap (Figure A, simplified). This system allows stimulation through “any” combination of electrodes whilst simultaneously recording EEG through non-stimulation electrodes, or with all electrodes used for recording immediately after stimulation (Figure B).

We will use the GTEN system to execute a “program” of brain stimulations (or “pings”), covering a spatially-distributed set of electrode configurations, and record the EEG responses following each stimulation (Figure C). We propose that clinically relevant information about brain state can be inferred from the properties of the set of responses evoked by the multi-dimensional program of stimulation. This approach would allow conscious states to be placed on a multi-dimensional landscape [2], rather than reduced to a uni-dimensional measure of conscious level [1].

A simple analogy is active sonar technology which, in its basic form, emits a well-defined pulse of sound, often called a "ping", listens for reflections (echoes) of the pulse, and draws inferences about nearby objects based on the properties of the reflected pulses. Our approach extends this analogy by systematically varying the spatiotemporal form of the stimulations (“pings”) in order to derive more information than would be possible from repetitions of a single form of stimulation. That is, a diverse program of stimulation reveals more facets of a system than that afforded by analysing responses to a single stimulation.

Question
This pilot project will investigate the question of whether EEG responses to diverse TES montages can even be distinguished in healthy resting awake participants, i.e. whether a machine learning classification algorithm can reliably classify EEG responses according to the stimulation applied. For example, given a set of 40 EEG recordings obtained from 10 repeats of 4 stimulation montages a-d, can the 10 EEG recordings corresponding to each of the montages a-d be correctly labelled a,b, etc. (Figure D)?

scott project diagram

Methods
We expect the student will carry out data collection in a small group of healthy participants using the GTEN system and analyse data using e.g. Matlab, signal processing, statistics and machine learning to answer the research question. The scope of the data acquisition and analyses experiments will depend on time, progress made and other experimental constraints.

Additional information

  • Existing source code will be available for some of the required tasks.
  • We expect students to take an active role in data curation and organisation.
  • We ask all students interested in the project to come and discuss the project in advance, having read any appropriate references.

Relevant references
Casali AG et al. A theoretically based index of consciousness independent of sensory processing and behavior. Science translational medicine. 2013;5(198):198ra
Bayne T, Hohwy J, Owen AM. Are there levels of consciousness? Trends Cogn Sci 2016;20:405–13

Computational modelling of neurovascular injury

Supervisors: Mazdak Ghajari (Design Engineering), David Sharp (Brain Sciences)

Mechanical forces produced in the brain during road traffic and sporting collisions, falls and assaults can damage the vessels in the brain. This leads to acute bleeding or blood brain barrier damage. Understanding how these forces cause injury is key to the design of prevention strategies or smart detection systems. We have developed a high fidelity computational model of traumatic brain injury (TBI), which allowed us to predict the location of pathology seen in post-mortem cases and MRI data from live patients. The model has been improved by incorporating detailed anatomy of vessels. This project will focus on using this model to simulate a large number of cases to predict the distribution of maximal forces in the vascular network. Neuroimages (SWI) from a large cohort of TBI patients will also be analysed to determine the patterns of vascular abnormalities. We will test whether there is a relationship between distribution of mechanical forces and injury patterns. The outcome of this study will be the validation of our computational model. This model will be a new tool that will allow us to test and improve the prevention effects of helmets.

Drug delivery across the blood-brain barrier using therapeutic acoustic wavelets

Supervisors: James Choi(Bioengineering), Magdalena Sastre (Brain Sciences)

Neurological diseases are among the most difficult diseases to treat, because drugs cannot enter the brain, because cerebral capillaries are lined by a blood-brain barrier (BBB).

We are developing a non-invasive and localised method of delivering drugs across the BBB. Using ultrasound and bubbles, we have been able to non-invasively and locally open the BBB for a short duration (less than 10 minutes) (Morse SM, et al., Radiology 2019), which allows drugs into the brain tissue.

The purpose of this project is to evaluate the effectiveness and safety of our acoustic wavelet drug delivery technology. For example, the student may explore the delivery of new and exciting drugs, such as nanobodies, antibodies, nanoparticles, etc; or may study the safety of the procedure, such as whether microglial cells and astrocytes have been activated.

This is a multidisciplinary project that spans physical acoustics, bubble physics, BBB structure and function, and neuroscience. Dr. James Choi will guide the use and development of the acoustic technology and the evaluation of drug delivery while Dr. Magdalena Sastre will guide the analysis of safety.

Dynamic Modulation of Circuits by Serotonin

Supervisors: Parastoo Hashemi (Bioengineering), Simon Schultz (Bioengineering)

Background and Rationale for Project
Serotonin has long thought to be neuromodulator, however the roles of this messenger in the modulation of brain activity remain poorly defined. The significance here is that serotonin is the primary target of most antidepressant compounds, that aim to increase extracellular serotonin levels. The rationale behind these agents is that extracellular serotonin levels are lower during depression (there remains, to date, no tangible evidence for this) and that these lower concentrations functionally change modulation of activity and lead to depression phenotypes. Because serotonin’s roles as a modulator are poorly defined and the chemical hypothesis of depression is yet to be verified, it is not surprising that antidepressants are not clinically efficacious for the vast majority of patients.

A fundamental and critical question, for better understanding how serotonin exerts its roles in the brain during normal function and disease, is how does serotonin modulate circuit activity? And do changes in this modulation alter circuit function?  There are very challenging questions to answer. Until recently, in vivo, it has been difficult to a) chemically measure serotonin and b) observe local waves of activity corresponding to volume transmission (the mode via which serotonin is thought to signal). The two PIs on this proposal are leaders in their respective fields in precisely these two types of measurements. Hashemi pioneered the first in vivo fast electrochemical methods to measure real time chemical changes in serotonin. Schultz has been one of the leaders of the development of multiphoton optical measurements of brain activity, and in the development of novel techniques for analysing optically and electrically recorded neurophysiological data.

In this interdisciplinary project, an MRes student will be tasked with combining these two niche methods in vivo to provide the first platform for studying serotonin’s roles in modulating circuit activity in the context of behavioural phenotypes.

Project Overview
The project is to marry fast scan cyclic voltammetry (FSCV)  at carbon fiber microelectrodes (CFMs) for serotonin measurements to in vivo multiphoton imaging of deep brain structures such as the hippocampus.

There are two specific objectives, outlined below:

Flexible, Implantable CFMs
Currently, FSCV experiments are performed with CFMs fabricated in glass pipettes. While these electrodes are successful for work in anesthetize animals, for our longer-term vision of studying neuromodulation during behaviour, these electrodes are not suitable because they are very fragile. The first phase of this project, thus, is to engineer a flexible material to insulate the carbon electrodes that can be utilized as an alternative to glass. The new devices, house in this new insulation material will be characterized rigorously for parity with their glass counterparts.

Simultaneous Measurements of Serotonin and Circuit Activity
A key part of the project will involve making measurements of serotonin in vivo while performing multiphoton imaging. This will involve (i) viral transduction of hippocampal CA1 neurons with a genetically encoded calcium indicator, using established techniques, (ii) optimising the surgery to allow both optical access to the hippocampus as well as insertion of a stimulation electrode as well as the carbon electrode and guidance to a location where it can be visualised the multiphoton imaging window, and (iii) performing two photon calcium imaging in the above experimental setup.

Expected Project and Mentoring Outcomes
At the completion of this project we expect to have a functional platform to test how serotonin modulates activity during behaviour with a view on how this modulation affects circuit function during depression, which is the basis of a PhD project.

The MRes student will receive unique mentoring and training in chemical measurements at ultra microelectrodes (Hashemi) and high resolution in vivo microscopic methods (Schultz) with the opportunity to utilize materials engineering and instrument optimization. There will be training in both fundamental and behavioural neuroscience.

Information theoretic data mining of large-scale neuropixels electrophysiological recordings from the mouse visual system

Supervisors: Simon Schultz (Bioengineering), Pier Luigi Dragotti (Electrical and Electronic Engineering)

The recent development of Neuropixels probes (comprising 960 electrode sites from which up to 384 selected sites can be read out) has substantially increased our ability to acquire large-scale, densely sampled neurons from the nervous system. With implantation of multiple Neuropixels arrays, simultaneous recordings can be made from many thousands of neurons. While the experimental techniques have advanced, the progress of data analysis techniques capable of taking advantage of the scaling of these recordings has lagged. The Allen Institute have recently made available a dataset comprising large scale Neuropixels recordings from the mouse visual system (see https://portal.brain-map.org/explore/circuits/visual-coding-neuropixels). This includes 6 Neuropixels probes, implanted such that they cover the dorsal lateral geniculate nucleus (dLGN), the lateral pulvinar nucleus (LP), and primary and secondary visual cortices. A battery of standard stimuli were applied. One disadvantage of the Allen dataset is the relatively crude receptive field (RF) mapping performed, in comparison to more specific dLGN studies such as Tang et al (2016). However, other stimuli such as natural scene images and movies were presented, which may present an opportunity to rectify this computationally. In this project we will firstly use a recently developed information-theoretic receptive field mapping approach which allows the recovery of RFs from natural movies, in order to recover detailed structure of dLGN RFs. Then, beginning with ON or OFF RFs in the dLGN, we will “mine” the dataset for overlapping RFs with the aim of reconstructing individual information processing operations performed throughout the circuit.

Prof Schultz will bring to this project expertise in systems neuroscience and information theoretic analysis of neurophysiological data. Prof Dragotti will bring expertise in sparse signal processing. Prof Nikolic developed the RF mapping algorithm we will use in the project, and will provide advice and assistance in its use, and further development. The project will provide multidisciplinary training in engineering approaches to the analysis of large-scale neuroscience datasets. It would suit a student with good python programming skills and a strong mathematical foundation (from electrical or biomedical engineering, physics or mathematics) and a strong desire to learn neuroscience.

Katz, M. L., Viney, T. J., & Nikolic, K. (2016). Receptive field vectors of genetically-identified retinal ganglion cells reveal cell-type-dependent visual functions. PloS one11(2).
Tang, J., Jimenez, S. C. A., Chakraborty, S., & Schultz, S. R. (2016). Visual receptive field properties of neurons in the mouse lateral geniculate nucleus. PloS one11(1).

Mapping the geometry of the “neural manifold” across the primate sensorimotor system

Supervisors: Juan Álvaro Gallego (Bioengineering), Claudia Clopath (Bioengineering)

Recent studies of neural population activity during behaviour have consistently uncovered that it is dominated by a handful of —around ten— neural covariation patterns (Gallego et al., 2017; Shenoy et al., 2013). These patterns span a neural manifold, a low-dimensional surface in the high-dimensional space defined by the activity of all recorded neurons. We refer to the neural population activity within the manifold as latent dynamics.

Remarkably, studying the neural manifold and its associated latent dynamics has helped advance our understanding of how the brain controls behaviour, providing insight into questions that had remained elusive when studying single neuron activity. The “manifold approach” has also helped made brain-computer interfaces (BCI) more robust (Gallego, Perich et al., 2020; Pandarinath et al., 2018).

Yet, there are many open questions regarding the geometrical properties of the neural manifold, with important implications both in basic neuroscience and BCIs. The goal of this project is to perform the first systematic comparison of manifold properties across the main areas of the sensorimotor system: premotor cortex, primary motor cortex, and primary sensory cortex. Such comparison will be carried out in a dataset with simultaneous recordings of around a hundred neurons in each of these areas, obtained as monkeys performed the same reaching task —the dataset from (Gallego, Perich et al., 2020). To assess the geometry of the neural manifold, we will fit the neural data with a broad range of manifold models that incorporate different assumptions about the data: from principal component analysis, and factor analysis, to isomap, and autoencoder neural networks. We will then explore the ability to predict the monkey’s behaviour from each of these different manifolds as we vary their assumed dimensionality. We expect this comparative study to inform about the difference in computations performed across these areas during a given behaviour, and on how to improve BCI design.

Dr. Gallego will bring to this project expertise in systems neuroscience, and neural data analysis, in particular focusing on the application of neural manifold approaches. Prof. Clopath is a leading expert in computational neuroscience and the application of mathematical methods and modelling tools to understand neural computations. This project would suit a student with good Python programming skills and a strong mathematical foundation (from electrical or biomedical engineering, physics or mathematics) and a strong desire to learn neuroscience.

References
J.A. Gallego, M.G. Perich, L.E. Miller, S.A. Solla. Neural manifolds for the control of movement. Neuron 94(5):978–84, 2017
J.A. Gallego*,M.G.Perich*, R.H. Chowdhury, S.A. Solla, L.E. Miller. Long-term stability of cortical population dynamics underlying consistent behaviour. Nature Neuroscience 23:260-276, 2020.
C. Pandarinath, K.C. Ames, A.A. Russo, A. Farshchian, L.E. Miller, E.L. Dyer, J.C. Kao. Latent factors and dynamics in motor cortex and their application to brain-machine interfaces. Journal of Neuroscience 38(44):9390-9401, 2018.
K.V. Shenoy, M.T. Kaufman, M. Sahani, M.M. Churchland. A dynamical systems view of motor preparation: implications for neural prosthetic system design. Prog Brain Res. 192 33-58, 2011.

Model-based estimation of upper limb joint kinematics and kinetics using surface electromyography

Supervisors: Dario Farina (Bioengineering), Angela Kedgley (Bioengineering)

Electromyography (EMG) is one of the few methods that provide a window on muscle activity and hence muscle force production during functional movements. As one of its applications, research in myoelectric control has produced active prosthetic devices with multiple degrees-of-freedom (DOFs), paralleled by EMG-based control strategies.

The aim of this project is to evaluate contrasting approaches for estimating upper limb joint kinematics and kinetics using surface EMG. Two such approaches are mathematical modelling (e.g. neural networks) and multibody dynamic musculoskeletal modelling.

From Dr. Farina students will gain the expertise to design and perform experiments to measure and process surface EMG signals. They will also learn how to build mathematical models linking measured EMG and kinematics.

From Dr. Kedgley students will learn to design and perform experiments to measure and process human body kinematics using motion capture systems. Furthermore, they will acquire skills in multibody dynamics modelling and simulation of the human upper limb.

The candidates are required to have strength programming in at least one of Python or MATLAB. Some knowledge of machine learning and biomechanical analysis is desirable.

Supervisor(s)Project title
Claudia Clopath, Andrei Kozlov Development of long-range connections in auditory cortex
Martyn Boutelle, Pantelis Georgiou, Mark Wilson Neurochemical CMOS array – bedside assay of ionic and inflammatory marker from the human brain
Dario Farina, Etienne Burdet, Emmanuel Drakakis, Patrick Kaifosh (Cognescent) Surface Electromyography for Brain-Machine Interface Applications
Ravi Vaidyanathan, Alison McGregor, Hildur Einarsdóttir (Ossur), Ásgeir Alexandersson (Ossur) Sensory Motor Interface for Lower Extremity Robots (SMILER)
Dario Farina, Paul Bentley A clinically-viable brain-computer interface for inducing neuroplasticity for stroke rehabilitation
Nir Grossman, Bill Wisden, Paul Matthews Development of non-invasive deep brain stimulation technology
Adam Hampshire, Aldo Faisal, Rob Leech, Gregory Scott Whole-brain dynamics and higher cognitive processing in disorders of consciousness
Tobias Reichenbach, Etienne Burdet Engineering tactile signals to aid hearing in noisy background
Chris Rowlands, Paul Chadderton 3D-resolved optogenetic excitation using time-averaged speckle patterns
David Sharp, Nir Grossman, Adam Hampshire, Peter Hellyer Closed-loop, personalized brain stimulation intervention for impairment of cognitive control
Mengxing Tang, Mike Warner, Matthew Williams 3D ultrasound computed tomography of the brain
Simon Schultz, Mauricio Barahona Analysis of calcium signals recorded endoscopically from the rodent brain
Mengxing Tang, Mike Warner Ultrasound technologies for brain imaging and therapy
Simon Schultz, Amanda Foust Ensemble coding models in the LGN: an asymmetry between ON and OFF?
Emmanuel Drakakis, Dario Farina Modular Reconfigurable Low-Power Stimulators