PET segmentation challenge using a data management and processing infrastructure

Challenge day schedule

Challenge date: October 21, 2016

Duration: Half-day (afternoon)

  • 14:00 – Opening/welcome
  • 14:05 – Presentation of the challenge computational infrastructure
  • 14:30 – Evaluated algorithms teasers (10 minutes each)
    • Team 1: “Ant Colony Segmentation Approach for Heterogeneous Volume Delineation in PET”. A. Ouahabi, V. Jaouen, M. Hatt, D. Visvikis, H. Fayad
    • Team 2: “Image features for tumor segmentation in PET using Random Forest”. S. Lieu, X. Huang, L. Li, W. Lu, S. Tan
    • Team 3: “Hybrid edge and region-based deformable model for PET tumor segmentation”. V. Jaouen, M. Hatt, H. Fayad, C. Tauber, D. Visvikis
    • Team 4: “Machine learning methods for accurate delineation of tumors in PET images”. J. Czakon, F. Drapejkowski, G. Zurek, P. Giedziun, J. Zebrowski, W. Dyrka
  • 15:15 – Coffee break and poster session
  • 16:15 – Workshop evaluation results and analysis
  • 17:15 – Round table discussion and perspectives of the challenge
  • 17:55 – Closing remarks



No computational equipment will be necessary, poster board can be requested by participants.


Organization committee

The challenge is jointly organized by France Life Imaging (for methodological and technical resources) and the TG211 AAPM taskgroup, regarding evaluation strategies and data provision.  The organization will be based on two specific boards: a scientific/clinical advisory board and a technical board.


Scientific/clinical advisory board currently appointed:

  • Dimitris Visvikis (LaTIM, Brest, France)
  • Mathieu Hatt (LaTIM, Brest, France)
  • Assen Kirov (MSKCC, New-York, USA, also head of TG211)
  • Federico Turkheimer (King’s College, London, U.K.)


Technical board currently appointed:

  • Frederic Cervenansky (Université Claude Bernard, Lyon)
  • Tristan Glatard (CNRS, Lyon)
  • Michael Kain (FLI-IAM, Inria, Rennes)
  • Baptiste Laurent (FLI-IAM, Inserm, Brest)


Funding support

Funding is provided by France Life Imaging. Three engineers will be 100% dedicated for the management of this challenge.