München 2004 – scientific programme
Parts | Days | Selection | Search | Downloads | Help
ST: Strahlenphysik und Strahlenwirkung
ST 2: Biomedizinische Bildgebung II
ST 2.7: Fachvortrag
Monday, March 22, 2004, 15:30–15:45, HS 129
Monte Carlo based 3D iterative reconstruction for high resolution positron emission tomography — •Magdalena Rafezas1, Melanie Dietz1, Brygida Mosler1, Markus Poegl2, Alexandros Stamatakis2, and Sibylle Ziegler1 — 1Nuklearmedizinische Klinik und Poliklinik, Klinikum rechts der Isar der TU Muenchen, Ismaningerstr. 22, 81675 Muenchen — 2Institut fuer Informatik, TU Muenchen, Boltzmannstr. 3, 85748 Garching b. Muenchen
Aim: Our goal was to calculate the system matrix P for 3D iterative image reconstruction for the small animal PET scanner MADPET-II using Monte Carlo (MC) simulations. The final purpose was to reconstruct a large field-of-view without compromising spatial resolution. A further objective was to study the feasability of using a database for handling the simulated data and P. Methods: We described MADPET-II in MC code. To calulate P we simulated a radioactive cylinder filling the scanner. The symmetries of the system allowed us to improve the statistical quantity P (Psym). For reconstruction the algorithm MLEM was implemented. Results: A total of 2.2 x 10**8 simulated events gave rise to P having N (non-zero elements)= 1.6x10**8 (1.2 GB of disk space). For P sym, N=8.0x10**8 (1.9GB). the transaxial resolution (FWHM) was 1.0 mm (r=0 mm) and 2.0 mm at 30 mm. When using Psym, the latter was improved to 1.4 mm. The reconstructed images showed a good contrast recovery and good SNR. Conclusions: 3D statistical iterative reconstruction using a MC based martix is feasable. DB2 has proved to be very practical framework to manage the large amounts of MC simulated data. The use of symmetries increased the accuracy of P and improved image quality.