Positron Emission Tomography Synthesis

Positron emission tomography (PET) image synthesis aims to generate high-quality PET images from readily available data like low-dose PET scans or structural MRI, mitigating the limitations of traditional PET imaging, such as high cost, radiation exposure, and limited accessibility. Current research heavily utilizes deep learning models, including generative adversarial networks (GANs) and diffusion models, often incorporating additional constraints like functional imaging information or anatomical segmentation to improve image fidelity and accuracy. This research is significant because it promises to expand access to PET imaging for broader clinical applications and facilitate larger-scale studies by overcoming the limitations of traditional PET acquisition.