Study and optimization of positioning algorithms for monolithic PET detectors blocks

  • Garcia De Acilu P
  • Sarasola I
  • Canadas M
 et al. 
  • 12


    Mendeley users who have this article in their library.
  • 8


    Citations of this article.


We are developing a PET insert for existing MRI equipment to be used in clinical PET/MR studies of the human brain. The proposed scanner is based on annihilation gamma detection with monolithic blocks of cerium-doped lutetium yttrium orthosilicate (LYSO:Ce) coupled to magnetically-compatible avalanche photodiodes (APD) matrices. The light distribution generated on the LYSO:Ce block provides the impinging position of the 511 keV photons by means of a positioning algorithm. Several positioning methods, from the simplest Anger Logic to more sophisticate supervised-learning Neural Networks (NN), can be implemented to extract the incidence position of gammas directly from the APD signals. Finally, an optimal method based on a two-step Feed-Forward Neural Network has been selected. It allows us to reach a resolution at detector level of 2 mm, and acquire images of point sources using a first BrainPET prototype consisting of two monolithic blocks working in coincidence. Neural networks provide a straightforward positioning of the acquired data once they have been trained, however the training process is usually time-consuming. In order to obtain an efficient positioning method for the complete scanner it was necessary to find a training procedure that reduces the data acquisition and processing time without introducing a noticeable degradation of the spatial resolution. A grouping process and posterior selection of the training data have been done regarding the similitude of the light distribution of events which have one common incident coordinate (transversal or longitudinal). By doing this, the amount of training data can be reduced to about 5% of the initial number with a degradation of spatial resolution lower than 10%.

Author-supplied keywords

  • Data processing methods
  • Data reduction methods
  • Gamma camera, SPECT, PET PET/CT, coronary CT angiography (CTA)

Get free article suggestions today

Mendeley saves you time finding and organizing research

Sign up here
Already have an account ?Sign in

Find this document

Get full text


  • P. Garcia De Acilu

  • I. Sarasola

  • M. Canadas

  • R. Cuerdo

  • P. Rato Mendes

  • L. Romero

Cite this document

Choose a citation style from the tabs below

Save time finding and organizing research with Mendeley

Sign up for free