You are here : Home > Overview > Overview

AAIMME

Overview of the project

​​​​​​​​​​​

Published on 31 March 2026

​Motivation

One of the key challenges in PET imaging is achieving a high temporal resolution, which is crucial for improving the signal-to-noise ratio and enhancing image quality. Currently, the temporal resolution of PET detectors is a few hundred picoseconds, but the goal is to reach a few tens of picoseconds. Improved temporal resolution allows for more precise selection of coincidental events, reducing noise in the acquisition. It also enables the use of Time-of-Flight (ToF) computation, which provides an estimate of the localization of the e+e– annihilation along the emission line of both gamma rays. Achieving a Coincidence Time Resolution (CTR) of tens of picoseconds can result in a ToF resolution of millimeters for each event, similar to the spatial resolution of the best dedicated PET systems.​

The AAIMME project leverages the advancements made by the ClearMind project, which focuses on developing innovative gamma-ray detectors. ClearMind utilizes monolithic PWO crystals with optimized optical interfaces to achieve high temporal resolution below 100 picoseconds, enabling precise 3D spatial reconstruction of interaction vertices. The AAIMME project builds upon these detectors to develop advanced AI methods for signal processing and image reconstruction, aiming to enhance the quality and reliability of PET imaging. By integrating the ClearMind detectors with AI-based approaches, AAIMME seeks to improve the spatio-temporal resolution of PET images, reduce patient exposure to radiation, and provide more accurate diagnostic information. This synergy between ClearMind and AAIMME represents a significant step forward in the field of medical imaging, combining cutting-edge detector technology with state-of-the-art AI techniques to improve PET imaging and patient care.


Key objectives

The AAIMME project is structured around three key objectives.

Detector Signal Processing

  • Develo​p Machine Learning (ML) approaches combined with expert signal pre-processing techniques to obtain reliable and quantified predictions of gamma interaction parameters.
  • ​Ensure that parameter reconstructions are accompanied by estimates of the uncertainties associated with the prediction, which can be taken into account during the PET imaging phase.

  1. ​Image Reconstruction

  • ​Develop im​​aging algo​rithms using the reconstructed interaction parameters as inputs.
  • ​Adapt the i​mage ​reconstruction process to achieve the expected gains in image quality.
  • ​Provide a reliable estimate of the uncertainties in the output parameters of the reconstructed image to enhance the reliability of t​heir interpretation.

  1. Validation on Experi​mental Benches

  • ​Evaluate the performance of the developed approaches by testing both on realistic Monte Carlo simulations and on experimental data.
  • Validate the entire chain: detection, detector signal processing, and image reconstruction; using dedicated Monte Carlo simulations and d​ata acquired on the tomXgam mechanical bench.


Impact

The expected impact of the AAIMME project on PET imaging is far-reaching, primarily in terms of achieving more precise temporal and spatial resolution. The improved resolution contributes to better image quality and a higher signal-to-noise ratio, enhancing image contrast. This advancement reduces the injected dose required to obtain images of the same quality as current standards, minimizing patient radiation exposure and facilitating earlier diagnostics and low-dose longitudinal studies.