W3 - Algorithm development

WG Leader: Bill Lionheart

This Working Group deals with the development of advanced tomographic reconstruction algorithms, as well as algorithms for other stages of the tomography pipeline (alignment, dynamic angle selection, etc.). As the specific details of the algorithms depend on the particular imaging technique (covered in the WGs T1-T3), the research in this WG focuses on generic algorithmic techniques that can be tailored for more specific applications in the other WGs.

The tasks for this Working Group include:
(a) Developing highly flexible and efficient implementations of iterative methods for linear inverse problems, such as ART, CGLS, LSQR.
(b) Developing and implementing flexible numerical schemes for nonlinear inverse problems (Gauss-Newton types methods, Levenberg-Marquardt, stochastic methods).
(c) Developing and implementing flexible algorithms for image reconstruction from highly limited data (Compressive sensing, Discrete tomography).
(d) Developing and implementing algorithms for real-time tomography (hybrid Filtered Backprojection methods, neural networks).
(e) Developing and implementing algorithms for parameter optimization and estimation (alignment parameters, estimation of motion parameters, etc.).