Reconstructions with TOF for in-beam PET ENVISION WP2 Meeting Feburary 2, 2010, CERN Georgy Shakirin Center for Radiation Research in Oncology – OncoRay Medical Faculty C. G. Carus Technische Universität Dresden
OncoRay experience with in-beam PET Since 1997 – installation, technical and clinical support of the first in-beam PET scanner installed at GSI, Darmstadt Complex simulations of ion interactions, production and transport of positron emitters, production and annihilation of positrons Reconstruction algorithms for limited angle PET scanners Routine evaluation of in-beam PET clinical data produced at GSI
Pricnciple of in-beam PET Prescribed dose Measurement Application Prediction: Simulation + Reconstruction Expected β+-activity distribution Reconstruction Measured β+-activity distribution Comparison
In-beam PET at GSI, Darmstadt Properties of the scanner 8 × 4 position sensitive detectors per head (BGO) Number of measured events: 30 – 500 × 103 Effectivity: ~ 1.5 % Spatial resolution: 5 – 10 mm g1 12C Results of in-beam PET monitoring g2 1997-2008: 440 patients monitored Detection of density changes in irradiated volume or range changes (~ 15 %) Detection of deviations in irradiation positioning (~ 10 %)
In-beam PET imaging Frontal Transversal Sagittal Reference image Reconstruction
Evaluation of image quality RMSE: Contrast recovery for phantom: 1 4 ? 1 ? 11, 13 14, 18.5 Radiuses of cold inserts, mm 3, 5, 7, 9 5, 6.5, 8.5, 11 Radiuses of hot inserts, mm 45 90 Radius, mm Height, mm Eigenschaft Standard 50 Small
Evaluation of image quality Modification of maximum particle range Normal range Overrange + 6 mm in water Underrange - 6 mm in water
Simulation workflow
TOF simulation Simplified geometrical simulation d1 d2 No interaction in detectors No scatter and attenuation List mode data: [detector1_id, detector2_id, t]
Reconstruction algorithms with TOF Time resolution lower than 0.2 ns (FWHM) Integration of TOF in iterative algorithms Shakirin et al., Proc. IEEE NSS MIC, Honolulu, USA, 4395-6, 2007 Time resolution higher than 0.2 ns (FWHM) Direct TOF method – reconstruction in real time Crespo, Shakirin et al., Phys. Med. Biol. 52, 6795-811, 2007
Reconstruction results In-beam PET scanner geometry similar to GSI installation OS-EM, no TOF OS-EM, TOF 1.2 ns DirektTOF 0.2 ns DirektTOF 0.1ns
Reconstruction results Reference image Direct TOF reconstruction, 0.2 ns OS-EM reconstruction, no TOF OS-EM reconstruction, TOF 1.2 ns
Conclusions Plans within WP TOF information increases the image quality for dual-head scanners even in case of low time resolution (1.2 ns) If TOF resolution is better than 0.2 ns, the Direct TOF method can be applied and reconstruction can be performed in real time Plans within WP Precise TOF simulations easily applicable to different geometries / detector types / DAQs of in-beam PET scanner Finding the best detector arrangement for TOF PET scanner - Further development of reconstruction algorithms for TOF PET