Uncertainty Quantification in Digital Breast Tomosynthesis

Applications are invited for a postgraduate research position leading to a PhD degree in Electrical Engineering in the Institute for Digital Communications within the School of Engineering at the University of Edinburgh.

The project is aimed at developing a practical, computationally efficient and diagnostically useful, uncertainty quantification algorithm for Digital Breast Tomosynthesis (DBT), exploiting recent advances in data-centric engineering and statistical model parameter fitting. DBT like any other limited angle tomography modality suffers from an inherent shortage of information that renders the image reconstruction problem as a statistical estimation problem with limited data. In the context of Bayesian estimation we seek to find the ‘family’ of (many) images that fit the observed data equally well within the error margins and are consistent with some available a priori anatomical hypothesis. While conventional deterministic inversion, e.g. filtered back projection, will brute force a selection of one image among the many plausible ones, the statistical approach tends to be more informative, as it investigates the relationship among the features of the whole family of likely images. In this way, while the deterministic approach may miss an important image feature (e.g. a small tumour) simply because it was not included in the one image selected, the statistical one computes the image (posterior expectation) and its uncertainty by integrating - averaging the likely images, e.g. tracing the conditional mean estimator and its covariance. For such method to be effective, it is necessary that the posterior distribution is sufficiently and efficiently sampled to bring about all likely images, and thus specially designed quadrature methods need to be developed that outperform Markov chain Monte Carlo in convergence rates. The research will require modelling the DBT data, coding the imaging algorithms, image reconstruction with clinical data, and mathematical analysis of the methods involved.

Further Information: 

This project is in collaboration with cancer research specialists, radiologists and image processing scientists. More information about the principles of tomosynthesis and limited angle tomography can be found in the references below

  1. F. J. Gilbert et al., Digital breast tomosynthesis (DBT): a review of the evidence for use as a screening tool, Clinical Radiology, 71, 2016.
  2. I. Sechopoulos, A review of breast tomosynthesis. Part II. Image reconstruction, processing and analysis, and advanced applications, Medical Physics, 40, 2013.

Closing Date: 

Sunday, December 31, 2017

Principal Supervisor: 


A first class Honours degree (or International equivalent) in engineering, physics, informatics or applied mathematics, ideally supplemented by an MSc Degree. 

Further information on English language requirements for EU/Overseas applicants.


Competitive funding subject to availability.

Further information and other funding options.