A Review of Deep-Learning-Based Approaches for Attenuation Correction in Positron Emission Tomography

Attenuation correction (AC) is essential for the generation of artifact-free and quantitatively accurate positron emission tomography (PET) images. PET AC based on computed (CT) frequently results in artifacts attenuation-corrected images, these mainly originate from CT PET-CT mismatches. The combined with a magnetic resonance imaging (MRI) scanner (PET/MRI) more complex than PET/CT, given that MR images do not provide direct information high-energy photon attenuation. Deep-learning (DL)-based methods improvement have received significant research attention as alternatives to conventional methods. Many DL studies were focused transformation into synthetic pseudo-CT or attenuation maps. Alternative approaches are dependent anatomical (CT MRI) can overcome limitations related current CT- MRI-based ACs allow quantification stand-alone scanners realization low radiation doses. In this article, review presented dual-modality PET/CT PET/MRI scanners, addition status progress DL-based approaches, improved performance AC.