Beca de estudios de doctorado en Brest, Francia: reconstrucción de imagen en la tomografía por emisión de positrones (PET)


PhD fellowship on tomographic image reconstruction for new detector architectures in PET imaging
Thesis location: Laboratory of Medical Information Processing (LaTIM), French Institute of Health and Medical Research (INSERM UMR1101), Brest, France.
Thesis supervisors: Julien Bert,; Dimitris Visvikis,
Period: 3 years, starting on September/October 2017

Context and objectives:
The development of new detectors in PET imaging, with the introduction of new semiconductor-based photodetectors, form the basis of future generations of clinical multimodal imaging devices. These detectors allow the measurement of additional and more detailed information, such as for example improved temporal resolution, depth of interaction, etc. This in turn presents opportunities for the development of novel reconstruction algorithms incorporating such details in combination with accounting for other detector specific parameters degrading overall image quality such as inter- and intra-crystal scatter. The incorporation of such new information in combination with the use of statistical model based approaches often requires enhanced computational power, in particular since the only efficient way to model and account for some detector specific parameters without any approximations is to use Monte Carlo (MC) simulations. Finally, the combination of detector corrections with that of patient specific phenomena such as attenuation, scatter and motion (respiration, cardiac) increase further the complexity and computational power needed in the reconstruction process.

In the last three years, we have developed a research program for the implementation of Monte Carlo-based numerical simulations on hybrid CPU/GPU architectures for medical applications (imaging and radiotherapy) [1]. At the same time, we have developed innovative methodology for modelling detector specific effects within the reconstruction process using a new multiline projector called IRIS (iterative random IDRF sampling) [2]. This projector has subsequently demonstrated superior performance in terms of resolution recovery, image contrast
and noise, compared to the use of traditional system matrix in the reconstruction process [3]. The main objective of this thesis will be to adapt the IRIS projector for different novel detector configurations such as for example the use of monolithic crystals, and the availability of time-of-flight and depth of interaction information. Comparison studies with other current state of the art solutions for such detector architectures will be considered. This work will be performed within the newly developed framework of the open-access CASToR reconstruction platform (

[1] “Geant4-based Monte Carlo simulations on GPU for medical applications” Bert J, Perez-Ponce H, El Bitar Z, Jan S, Boursier Y, Vintache D, Bonissent A, Morel C, Brasse D, Visvikis D. Phys Med Biol. 2013 Aug 21;58(16):5593-611.
[2] “Fully 3D PET List-Mode reconstruction includingan accurate detector modeling on GPU architecture” Awen Autret, Julien Bert, Olivier Strauss and Dimitris Visvikis 12th International meeting on Fully 3D Image Reconstruction in Radiology and Nuclear Medicine, 2013
[3] “Detector modeling in PET list-mode reconstruction: Comparison between pre-calculated and on-the-fly computed system matrix” Awen Autret, Matthieu Moreau, Thomas Carlier, Julien Bert, Olivier Strauss, Dimitris Visvikis, IEEE Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC) proceedings, 2015, Pages: 1 - 3, DOI: 10.1109/NSSMIC.2015.7582100

Education: The candidate must have a Master's degree in Signal/Image processing, Electrical/Electronic/Biomedical Engineering, Physics, Computer Science or Applied Mathematics.
Scientific interests: Medical imaging, tomographic reconstruction, numerical simulation.
Programming skills: C/C++, CUDA (optional).
Languages: English (knowledge of French is not required)

If you require any further information prior to application, please contact Prof. Dimitris Visvikis
To apply send CV, grades/marks and a brief statement of interest by email as soon as possible and preferably before the end of April 2017 to:
Dimitris Visvikis, and Julien BERT,