Skip to main content
Fig. 5 | EJNMMI Research

Fig. 5

From: ParaPET: non-invasive deep learning method for direct parametric brain PET reconstruction using histoimages

Fig. 5

Qualitative and quantitative comparison of results from an example subject. Shown are a images generated at steps a, b, c, e’ and g in the reconstruction pipeline (refer to Fig. 1), where PSNR and SSIM were referenced to TOF-OSEM, b (left to right) net influx rate constant (Ki) reconstructed using the DL method, NLS without and with constraints (Ki-NLS and Ki-NLS-C) and Patlak method (Ki-Patlak)

Back to article page