(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose: Simultaneous dual-isotopes PET imaging (DIPET) could provide two different body functions, perfectly registered information and reduced scan time. However, because signals used for image reconstruction by PET for both isotopes are from the 511 keV annihilation photons, it is difficult to separate signals from the two isotopes. In this work, we proposed a novel separation method for DIPET based on the triple coincidence and artificial neural network.
Methods: The DIPET method is based on the use of a combination of a pure and a nonpure positron emitter for dual isotopes PET proposed by the University of British Columbia (UBC). We improved this method by introducing an artificial neural network for further enhancing the image quality. To validate the proposed method, simulations of a three-rod phantom and a MOBY phantom (filled in F-18 and I-124) is performed using GATE/MPHG software.
Results: For the F-18 image quantitative analysis in different activities, the NRMSEs of the predicted image using the proposed method were 0.0593, 0.0442 and 0.0224 while 0.1923, 0.1400 and 0.0702 for UBC's method. For the I-124 image quantitative analysis in different activities, the NRMSEs of the predicted image using the proposed method are 0.0871, 0.0845 and 0.1127 while b 0.2849, 0.2666 and 0.3906 for UBC’s method.
Conclusion: Our proposed method could get a higher recovering rate of non-pure emitters while retaining similar quantitative recovery with the single-isotope image. Moreover, the method can be applied to other pure emitters and non-pure emitters.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
