AAIMME - Apprentissage Automatique pour l'Imagerie Moléculaire et la MÉdecine du futur ​

Welcome to the AAIMME project website! 

Our mission is to improve Positron Emission Tomography (PET) systems through innovative Artificial Intelligence (AI) techniques, enhancing image quality and reducing patient exposure to radiation.

The AAIMME project aims to develop reliable AI tools that optimize the ultra-fast detection chain and image reconstruction processes. By leveraging advanced AI methods, we strive to improve the precision of diagnostics and therapeutic monitoring for patients.

Our objectives include developing AI methods to assess uncertainties from acquisition to the reconstructed image, providing effective support for healthcare staff in their decision-making. The methodology will be validated on experimental data representative of PET acquisitions.

The project involves multidisciplinary research, combining expertise in AI methods, medical imaging physics, and instrumental development physics.

The AAIMME project has reveived funding from the French National Research Agency, N°ANR-24-CE45-7518.

ReportAAIMME gallery
  • View in full screen
  • Principe ClearMind
    Principe ClearMindView in full screen
  • Reconstruction ML position
    Reconstruction ML positionView in full screen
  • Banc tomXgam
    Banc tomXgamView in full screen
  • Principe ClearMind
  • Reconstruction ML position
  • Banc tomXgam

​Positron Emissi​​​on Tomography (PET) is a functional nuclear medicine imaging modality that involves the administration of a radiotracer labeled with a positron emitter to the patient. This radiotracer targets a specific molecular pathway, and the coincidental detection of gamma rays pairs emitted 180° apart by positron-electron (e+e–) annihilations within the organs to be imaged is exploited to create a functional image. The statistical accumulation of such events is then used to image the bioactivity distribution of the chosen tracer inside the human body, providing crucial information for diagnosis and therapeutic follow-up.

The reliability of a PET diagnosis depends significantly on the quality of the reconstructed image, particularly its signal-to-noise ratio. This ratio is directly influenced by the activity injected into the patient. While greater activity can improve the signal-to-noise ratio, it is equally important to minimize the injected activity to reduce the patient's exposure to radiation and the risk of damaging healthy cells or organs. Additionally, reducing the injected dose can make PET exams more cost-effective and accessible.