Over the past three years the PET camera installed at the tumour therapy facility in GSI has proven its capability of contributing to the quality assurance of ¹²C radiotherapy [1]. In this report, we present the first studies concerning the PET camera design and implementation onto a dedicated hospital-based ion beam facility for cancer therapy, proposed to be built in Heidelberg [2] due to the promising results achieved with the GSI pilot project [3]. In the proposed facility, due to the presence of a rotating gantry which will deliver the heavy ion beam - thus satisfying an important medical need [4], the flexibility of the PET camera must be enhanced so that it does not collide with the patient or the couch nor with the beam gantry, as well as it allows a fast access of the physicians to the patient.
Figures 1 and 2 depict possible PET scanner implementations. In Fig. 1, the scanner moves along the patient couch and, thus, can be positioned around the region being irradiated. An aperture on the tomograph ring allows the beam to pass through without touching the g-ray detectors. This configuration can provide a full coverage of the volume under observation if the scanner rotates 180^° around its axial direction (Z_PET) during the beam extraction cycle (spill off, ~  2 s). This detail, besides being relevant for the image quality, is also important for a quantitative analysis of the measured b⁺-activity. If typical dimensions of a PET scanner are applied, an aperture of 30 cm for the beam is considered (param. 7) and the distance between the scanner and its support ring is 40 cm (param. 10), the map plotted in Fig. 3 is obtained. The black and dark grey areas correspond to beam gantry and patient couch angle combinations not suitable for therapy because the beam penetrates the patient through the trunk of the body (caudo-cranial direction). The angle combinations mapped in light grey are free for irradiation only if the beam leaving the target volume does not activate substantially the scanner support structure. In this configuration the scanner never approaches the patient.

Fig. 1 PET scanner connected with the patient couch. The couch lies along Z_PET and the camera rotates, before the beginning of the treatment, facing the beam with its aperture (not shown). Legend: 1) Nozzle distance to isocenter, 2) Beam radius, 3) Nozzle radius, 4) Scanner outer radius, 5) Scanner inner radius, 6) Scanner width, 7) Scanner aperture for beam, 8) Support ring outer radius, 9) Support ring width and 10) Distance between scanner and its support ring.

Fig. 3 Collision study for the case PET connected with the patient couch. Black area: physical collision between beam gantry and scanner. Dark grey area: beam collides with scanner support ring before the target volume. Light grey area: as dark grey but after irradiating the target volume.

In Fig. 2 the PET scanner is depicted, which is assumed to be connected to the gantry and to be placed parallel to the beam nozzle (the beam does not collide with the scanner or its support structure in any situation). A collision study yields a minimal scanner aperture of 75 cm in order for no physical collisions to occur between scanner and patient/couch. If the scanner is placed parallel to the beam nozzle the minimum aperture needed decreases to 65 cm (the present dual-head PET scanner at the GSI therapy facility has 63 cm aperture). These collision studies did not take into account the movement of the scanner into the measurement position.
In addition to the collision studies summarized, we have developed the tools to quantify the spatial resolution degradation as one moves from a closed ring to an open ring PET camera configuration: (i) a simulation capable of treating several camera geometries and (ii) a flexible image reconstruction routine being capable of reading the output from the simulations. The reconstruction uses an iterative procedure based on the maximum likelihood estimation maximization algorithm. Due to the enormous amount of crystal combination possibilities (over 150 million), dynamic memory allocation is used in conjunction with a developed factorization scheme, which obliged the routine to differ substantially from the one presently used at the GSI therapy unit [5].
In Fig. 4 we depict the first results on the spatial resolution degradation if one moves from a closed ring to an open ring detector assembly. No Compton scattering effects on the detector were simulated yet, the attenuation of the g-rays in the crystals and thus the depth-of-interaction influence on the spatial resolution was taken into account. The degradation in spatial information experimentally observed for detector crystals coupled to photomultipliers according to the modified Anger principle was also not included.

Fig. 4 Comparison between the spatial resolution achieved with a closed ring versus an open ring positron camera (diameter: 82.3 cm, aperture: 63 cm, Fig. 1), built up of crystals of 5.0 × 4.5 mm² frontal surface and 30 mm depth.

¹ Gesellschaft für Schwerionenforschung Darmstadt

References

[1]	W. Enghardt et al., GSI Scientific Report (1999) 164-5
[2]	K.D. Gross, M. Pavlovic (eds.), Proposal for a dedicated ion beam facility
	for cancer therapy, GSI Darmstadt, 1998
[3]	J. Debus et al., Strahlenther. Onkol., 176 Nr 5, (2000) 211-6
[4]	O. Jäkel and J. Debus, Phys. Med. Biol. 45 (2000) 1229-41
[5]	K. Lauckner, Ph.D. Thesis, Dresden University of Technology, 1999

IKH 06/27/01 © P. Crespo