Publikationsrepositorium - Helmholtz-Zentrum Dresden-Rossendorf

Purpose: To investigate the possibility of detecting patient mispositioning in carbon-ion therapy with particle therapy PET in an automated image registration based manner.
Methods: Tumors in the head and neck (H&N), pelvic, lung and brain region were investigated. Biologically optimized carbon ion treatment plans were created with TRiP98. From this treatment plans the reference +-activity distributions were calculated using a Monte Carlo simulation. Setup errors were simulated by shifting or rotating the CT. The expected + activity was calculated for each plan with shifts. Finally, the reference particle therapy PET images were compared to the "shifted" +-activity distribution simulations using the Pearson's correlation coecient (PCC). To account for dierent PET monitoring options the in-beam PET was compared to three dierent in-room scenarios. Additionally the dosimetric eects of the CT misalignments were investigated.
Results: The automated PCC detection of patient mispositioning was possible in the investigated indications for cranio-caudal shifts of 4 mm and more, except for prostate tumors. In the rather homogeneous pelvic region the generated +-activity distribution of the reference and compared PET image were to much alike. Thus setup errors in this region could not be detected. Regarding lung lesions the detection strongly depended on the exact tumor location: in the center of the lung tumor misalignments could be detected down to 2 mm shifts while resolving shifts of tumors close to the thoracic wall was more challenging. Rotational shifts in the H&N and lung region of +6 and more could be detected using in-room PET and partly using in-beam PET. Comparing in-room PET to in-beam PET no obvious trend was found. However, among the in-room scenarios a longer measurement time was found to be advantageous.
Conclusion: This study scopes the use of various particle therapy PET verication techniques in four indications. The automated detection of patients' setup errors was investigated in a broad accumulation of data sets. The evaluation of introduced setup errors is performed automatically, which is of utmost importance to introduce highly required particle therapy monitoring devices into the clinical routine.