PhD Studentship: Joint Reconstruction-segmentation for Imaging

A crucial image processing task is image segmentation: we often seek to identify the presence and location of key features in an image, for example a tumour in a medical scan. But in practice, we do not observe images directly. An image observed by, for example, an MRI or CT scanner, must be reconstructed from measurements which are often damaged, incomplete, and/or corrupted by noise. In practice, therefore, whenever one seeks to solve a segmentation problem, one really needs to solve both a reconstruction and a segmentation problem. Joint reconstruction-segmentation is a technique which solves both of these tasks together, using each to guide the other. This idea of task-adapted reconstruction has seen considerable mathematical attention in recent years.

This project may go in a broad constellation of directions. One direction might be to make the techniques in this pipeline more sophisticated, building on current methods by incorporating new techniques. Another might be to develop a more developed statistical foundation for this approach, potentially opening doors to novel approaches. Another direction is to go beyond segmentation. Often in practice a segmentation is only a means to a further end, such as a prognosis: can we rigorously extend the joint pipeline to those downstream tasks? This project will build on work in collaboration with the Memorial Sloan Kettering Cancer Center.

Funding notes:

Funding is available through the School of Mathematics for a suitably strong candidate.

The scholarship will cover tuition fees, training support, and a stipend at standard rates for 3-3.5 years;

Candidates are encouraged to make an informal inquiry with Dr Jeremy Budd (j.m.budd@bham.ac.uk).

For application details, please click the above “Apply” button.

J. Adler, S. Lunz, O. Verdier, C.-B. Schönlieb, and O. Öktem, “Task adapted reconstruction for inverse problems”, Inverse Problems 38 (2022), p. 075006, https://doi.org/10.1088/1361-6420/ac28ec.

V. Corona, M. Benning, M. J. Ehrhardt, L. F. Gladden, R. Mair, A. Reci, A. J. Sederman, S. Reichelt, and C.-B. Schönlieb, “Enhancing joint reconstruction and segmentation with non-convex Bregman iteration”, Inverse Problems 35 (2019), p. 055001, https://doi.org/10.1088/1361-6420/ab0b77.

J. M. Budd, Y. van Gennip, J. Latz, S. Parisotto, and C.-B. Schönlieb, “Joint reconstruction-segmentation on graphs", SIAM Journal on Imaging Sciences 16 (2023), p. 911-947, https://doi.org/10.1137/22M151546X.

Z. Wu, T. Yin, Y. Sun, R. Frost, A. van der Kouwe, A. V. Dalca, and K. L. Bouman. "Learning task-specific strategies for accelerated MRI." IEEE Transactions on Computational Imaging (2024), https://doi.org/10.1109/TCI.2024.3410521. 

Advert information