@article{WedrowskiBruyndonckxTavernieretal.2009,
  author    = {Wedrowski, M. and Bruyndonckx, P. and Tavernier, S. and Zhi, L. and Dang, J. and Mendes, P. R. and Perez, J. M. and Ziemons, Karl},
  title     = {Robustness of neural networks algorithm for gamma detection in monolithic block detector, positron emission tomography},
  series = {2009 IEEE Nuclear Science Symposium Conference Record (NSS/MIC)},
  journal   = {2009 IEEE Nuclear Science Symposium Conference Record (NSS/MIC)},
  isbn      = {1082-3654},
  pages     = {2625 -- 2628},
  year      = {2009},
  abstract  = {The monolithic scintillator block approach for gamma detection in the Positron Emission Tomography (PET) avoids estimating Depth of Interaction (DOI), reduces dead zones in detector and diminishes costs of detector production. Neural Networks (NN) are very efficient to determine the entrance point of a gamma incident on a scintillator block. This paper presents results on the robustness of the spatial resolution as a function of the random fraction in the data, temperature and HV fluctuations. This is important when implementing the method in a real scanner. Measurements were done with two Hamamatsu S8550 APD arrays, glued on a 20 {\~A}— 20 {\~A}— 10 mm3 monolithic LSO crystal block.},
  language  = {en}
}