|Funding for:||UK Students, EU Students|
|Funding amount:||£14,777 Annual stipend (2018/9 rate) + Home/EU tuition fees + training support grant|
|Placed On:||17th October 2018|
|Closes:||30th November 2018|
Lead supervisor: Dr Mohammad Golbabaee, Department of Computer Science, University of Bath, email: firstname.lastname@example.org
This project is one of a number that are in competition for Research Council funding.
Full Project Description
With recent advances in machine learning, we are witnessing a new wave of data-driven approaches emerge to uncover the physical models behind many complex systems in modern applications. In this project, we particularly aim at developing machine learning techniques for the Quantitative Magnetic Resonance Imaging (Q-MRI). In Q-MRI rather than simply forming a single MR image with only contrast information (i.e. qualitative MRI), physicists are interested in uncovering the NMR properties of tissues such as the relaxation times and proton density from a time-series of images. The main challenge of standard Q-MRI approaches is their very long acquisition time. Recently, MR Fingerprinting (MRF) emerged to address this issue using combined tools from Compressed Sensing theory and pattern recognition. The MRF framework simulates offline a dictionary of fingerprints (i.e. magnetic responses) for all NMR parameters and searches through this dictionary to track back the underlying parameters from the measurements. However this approach is known to be non-scalable and bottlenecked by the heavy storage and computation requirements of searching through very large dictionaries in multi-parametric Q-MRI applications.
Combined with recent advances in Deep Learning models, this project aims at building dictionary-free systems to address the non-scalability issue of the MRF problem. This requires designing a new neural network architecture and an efficient training strategy to learn complex physical dynamics behind Q-MRI. This project will run in close collaboration with one of the world-leading medical imaging industries and the proposed methodologies will be validated against real-world medical imaging datasets.
Research Council funding is available on a competition basis to Home and EU students who have been resident in the UK for 3 years prior to the start of the project.
Funding will cover Home/EU tuition fees, a stipend (£14,777 per annum for 2018/9) and a training support fee of £1,000 per annum for 3.5 years.
Applicants classed as Overseas for tuition fee purposes are NOT eligible for funding; however, we welcome all-year-round applications from self-funded candidates and candidates who can source their own funding.
Applicants should hold, or expect to receive, a First Class or good Upper Second Class Honours degree, or the equivalent from an overseas university. A master’s level qualification would also be advantageous.
Formal applications should be made via the University of Bath’s online application form for a PhD in Computer Science: https://samis.bath.ac.uk/urd/sits.urd/run/siw_ipp_lgn.login?process=siw_ipp_app&code1=RDUCM-FP01&code2=0013
Preferred start date: 21 January 2019. We are willing to discuss an alternative start date but candidates must be prepared to start by 1 April 2019 at the latest.
NOTE: We are hoping to fill this position as quickly as possible and applications may close earlier that the advertised deadline if a suitable candidate is found; early application is therefore recommended.
Type / Role: