Fig. 3
From: Applications of artificial intelligence and deep learning in molecular imaging and radiotherapy
Synthetic CT generation from T1-weighted MR image using a generative adversarial network. a Target MRI, b reference CT, c atlas-based, d deep learning-based, and e segmentation-based synthetic CT images. PET ages corrected for attenuation using f reference CT, g atlas based, h deep learning based, and i segmentation-based synthetic CT images. PET bias map (relative error in %) for j atlas based, k deep-learning based, and l segmentation-based attenuation correction