Uniform acquisition modelling across PET imaging systems: Unified scatter modelling
View / Open Files
Publication Date
2016Journal Title
2016 IEEE Nuclear Science Symposium, Medical Imaging Conference and Room-Temperature Semiconductor Detector Workshop, NSS/MIC/RTSD 2016
Conference Name
Nuclear Science Symposium, Medical Imaging Conference and Room-Temperature Semiconductor Detector Workshop (NSS/MIC/RTSD), 2016
ISSN
1095-7863
ISBN
9781509016426
Volume
2017-January
Type
Conference Object
Metadata
Show full item recordCitation
Markiewicz, P., Ehrhardt, M., Atkinson, D., Arridge, S., Hutton, B., & Ourselin, S. (2016). Uniform acquisition modelling across PET imaging systems: Unified scatter modelling. 2016 IEEE Nuclear Science Symposium, Medical Imaging Conference and Room-Temperature Semiconductor Detector Workshop, NSS/MIC/RTSD 2016, 2017-January https://doi.org/10.1109/NSSMIC.2016.08069584
Abstract
© 2016 IEEE. PET imaging is an important tool commonly used for studying disease by research consortia which implement multi-centre studies to improve the statistical power of findings. The UK government launched the Dementias Platform UK to facilitate one of the world's largest dementia population study involving national centres equipped with state-of-the-art PET/MR scanners from two major vendors. However, the difference in PET detector technology between the two scanners involved makes the standardisation of data acquisition and image reconstruction necessary. We propose a new approach to PET acquisition system modelling across different PET systems and technologies, focusing in particular on unified scatter estimation across TOF (time-of-flight) and non-TOF PET systems. The proposed scatter modelling is fully 3D and voxel based, as opposed to the popular line-of-response driven methods. This means that for each emitting voxel an independent 3D scatter estimate is found, inherently preserving the necessary information for TOF calculations as well as accounting for the large axial field of view. With adequate sampling of the input images, the non-TOF scatter estimate is identical to the summed TOF estimates across TOF bins, without an additional computational cost used for the TOF estimation. The model is implemented using the latest NVIDA GPU CUDA platform, allowing finer sampling of image space which is more essential for accurate TOF modelling. The high accuracy of the proposed scatter model is validated using Monte Carlo simulations. The model is deployed in our stand-alone image reconstruction pipeline for the Biograph mMR scanner, demonstrating accurate 3D scatter estimates resulting in uniform reconstruction for a high statistics phantom scan.
Identifiers
External DOI: https://doi.org/10.1109/NSSMIC.2016.08069584
This record's URL: https://www.repository.cam.ac.uk/handle/1810/279223
Rights
Licence:
http://www.rioxx.net/licenses/all-rights-reserved
Statistics
Total file downloads (since January 2020). For more information on metrics see the
IRUS guide.