DOSIMETREYE: A GenAI-based Tool for 3D Dosimetry from 2D Imaging for Preclinical Studies

Short description of the innovation study

Small animal imaging and dosimetry prediction tools can enable the effective reduction of over 80% of animal sacrifice and drastically speed up the assessment of drug delivery studies in preclinical research. The vision of the ΔosimetrEYE project is to develop the first AI-based dosimetry tool that will provide absorbed dose analysis in real-time alongside imaging acquisition. This is envisaged as a highly innovative digital solution for dosimetry assessment during the medical imaging of radiopharmaceuticals, which could later be expanded for clinical purposes.

The innovation goes beyond the state-of-the-art, as advanced GenAI algorithms will generate 3D dose maps from 2D dynamic images and predict absorbed dose assessments at the organ level. We utilise: i) MC simulations to provide the ground truth, ii) AI techniques for generating and predicting absorbed doses per organ, iii) HPC resources to perform computationally demanding tasks, and iv) preclinical imaging data to feed the proposed GenAI training algorithms.

Organisations involved: 

End User: BIOEMTECH 

HPC expert: GRNET 

XAI expert: ALETHIA

GRNET is a member of the NCC Greece.