Scientific Output

Scientific aspects of NiftyPET

The core routines are written in CUDA C and embedded in Python C extensions to enable user-friendly and high-throughput executions on NVIDIA graphics processing units (GPU). The scientific aspects of this software platform are covered in two open-access publications:

Advanced imaging with the ACR phantom

Quantitative measurement of the spatial resolution using the measured knife-edge response \(K(x)\), for any phantom insert with different attenuation and activity properties, thus facilitating comprehensive and quantitative characterisation of PET/MR or (PET/CT) scanners in multisite clinical studies, e.g., in the Dementias Platform UK network.

P.J. Markiewicz, C. da Costa-Luis, J. Dickson, A. Barnes, G. Krokos, J. MacKewn, T. Clark, C. Wimberley, G. MacNaught, M.M. Yaqub, J.D. Gispert, B. F. Hutton, P. Marsden, A. Hammers, A.J. Reader, S. Ourselin, K. Herholz, J.C. Matthews, F. Barkhof. Advanced quantitative evaluation of PET systems using the ACR phantom and NiftyPET software. Med. Phys., accepted for publication.

MR-PET registration uncertainty analysis

Novel MR-PET registration uncertainty analysis, indicating that registration software has the biggest effect on MR-PET registration precision, followed by reconstruction parameters (i.e., iterations, smoothing) and PET count level. Although PVC can significantly improve the PET signal, it also increases PET signal variability since it relies on precise MR-PET registration. More details can be found in [4]:

P.J. Markiewicz, J.C. Matthews, J. Ashburner, D.M. Cash, D.L. Thomas, E. De Vita, A. Barnes, M.J. Cardoso, M. Modat, R. Brown, K. Thielemans, C. da Costa-Luis, I. Lopes Alves, J.D. Gispert, M.E. Schmidt, P. Marsden, A. Hammers, S. Ourselin, and F. Barkhof (2021). Uncertainty analysis of MR-PET image registration for precision neuro-PET imaging. Neuroimage 655 232, 117821. https://doi.org/10.1016/j.neuroimage.2021.117821

Fast PET image reconstruction

An example application of NiftyPET in the development of novel image reconstruction using advanced randomized optimization algorithms to solve the PET reconstruction problem for a very large class of non-smooth priors [5]:

M.J. Ehrhardt, P.J. Markiewicz, C-B. Schönlieb (2019). Faster PET reconstruction with non-smooth priors by randomization and preconditioning. Phys. Med. Biol. 64(22), https://doi.org/10.1088/1361-6560/ab3d07

Dynamic PET image reconstruction

Dynamic PET image reconstruction for reduced acquisition time PET pharmacokinetic modelling [6]:

C.J. Scott, J. Jiao, A. Melbourne, N. Burgos, D.M. Cash, E. De Vita, P.J. Markiewicz, A. O’Connor, D.L. Thomas, P.S.J. Weston, J.M. Schott, B.F. Hutton, S.Ourselin (2018) Reduced acquisition time PET pharmacokinetic modelling using simultaneous ASL–MRI: proof of concept. Journal of Cerebral Blood Flow & Metabolism. https://doi.org/10.1177/0271678X18797343