minor repositioning errors in connection with sharp density gradients in the beam path or changes in the physical condition of the patient. To quantify such deviations from PET data, improved attenuation, scatter  and random  correction methods for the measured data as well as refined models (positron range, photon scattering) for predicting the β-activity distribution from the treatment plan  are introduced. Furthermore, to compensate for the metabolic washout of the positron emitters, which is correlated to the local blood flow , the tissue dependent biological half-lives of the β-activity have to be known. It is expected that this information can be extracted from the more than 2600 list mode PET data sets measured during patient irradiations so far. An appropriate code that allows the data to be analyzed simultaneously in ordinary space and in the time domain has been developed. To increase the flexibility in treatment planning the therapy facility will be equipped with a chair  for irradiating patients in a sitting position. This required to build a completely new PET gantry (Fig. 2), which allows the detector heads to be rotated around the beam axis . Combining an in-beam PET scanner with an ion beam gantry, as it is planned for the Heidelberg clinical facility, requires new technical solutions, namely new scanner configurations. To predict their imaging properties a versatile PET simulation and reconstruction tool has been developed . Gantry-delivered multi-field irradiations are expected to result in PET scans of low counting statistics, and thus, an optimization of the signalto-noise ratio is required. For this we studied  the possibility of using the new scintillator material lutetium orthosilicate (LSO), which is superior to the currently used bismuth germanate (BGO). This feasibility study was addressed to the influence of background coincidences arising from the β-decay of Lu. Natural Lu contains 2.59 % of this isotope leading to Figure 2: The new PET gantry. (Photo: A. Zschau, GSI).