Denoising Approaches by Artificial Intelligence in PET MRI for clinical routine application

Motivation: Two big problems encountered in hybrid MR PET exams are: long duration of the exams and the optimization of the administered activity. Goal(s): I tried to evaluate the feasibility of decreasing the time and dose using an artificial intelligence tool in reconstruction while preserving the performance of PET and MR. Approach: By analyzing the literature, it was possible to identify the optimal reconstruction strategies for PET and MR imaging that utilize artificial intelligence to save dose and time. Results: Deep learning techniques have made significant advances in data reconstruction images from examinations with low scan times or radiopharmaceutical doseImpact: Artificial intelligence in MR PET is a promising approach. The impact on the health of patients is undeniable, especially in the paediatric population. This approach reduces the dose and consequently the cost of radiopharmaceuticals and increases productivity and efficiency.