At present the commissioning of the experimental heavy ion radiotherapy unit at the Gesellschaft für Schwerionenforschung (GSI) Darmstadt is being accomplished. Simultaneously we are investigating the capabilities and limitations of the PET system that has been installed at the medical treatment site for in-situ control of the tumour irradiations. In the following results from a series of experimental studies will be introduced.

One of the most essential problems in this context is the accurate determination of primary particle ranges from measured ß⁺-activity profiles. We studied this for ß⁺-emitter distributions obtained by means of irradiating homogeneous lucite blocks with monoenergetic ¹²C beams in the energy range between 91.5 and 306 AMeV. The ß⁺-activity depth distribution constructed from these PET data have been parameterized by an analytical model. Applying the same model to depth distributions of ß⁺-activity and dose obtained by a realistic Monte Carlo simulation of the interaction of the ¹²C beam with the target material allows the particle range to be determined with an accuracy of one millimetre at least.

Similar experiments have been performed to investigate the sensitivity of this PET imaging method for detecting a mispositioning of the dose distribution due to a too short or too long range of the ¹²C ions in tissue. These have shown that range differences down to 0.8 mm can be resolved.

The introduction of the PET monitoring of heavy ion therapy into clinical practice requires the combination of the PET measurements with all the other components of heavy ion therapy, in particular with treatment planning and dose delivery. We have established a procedure which results in images where X-ray computed tomograms of the tumour region are superimposed by the ß⁺-activity distributions obtained from the reconstruction of the measured data and from Monte Carlo simulation based upon the treatment plan and the dose delivery record. The comparison between the simulated and the measured PET images considering anatomical details reveals discrepancies between planned and applied dose. If necessary, the treatment plan can be modified before next therapy fraction. This procedure has been successfully tested by the irradiation of an Alderson head phantom. The expected ß⁺-emitter distributions obtained from simulation are in good agreement with the actually measured distributions.

The experiments show the benefit from the 3D in-situ PET measurement for the quality assurance of heavy ion tumour therapy. Further studies will focus on the influence of physiological processes (blood flow, breathing, metabolism) on dose localisation by PET.