Assessment of an open magnetic resonance (MR) system for radiotherapy treatment planning of extracerebral tumors

Purpose/Objective: Computed tomography (CT) has become the most widely used source of image data for 3D treatment planning in radiotherapy. Magnetic resonance imaging (MRI) provides superior soft-tissue contrast and multiplanar imaging capabilities and has therefore gained an important role in target volume definition and delineation of organs at risk, especially for CNS tumors. However, MRI data do not provide the electron density of the respective tissue and the geometric reliability is inferior to CT. The purpose of this project was to establish an open MR system for radiotherapy treatment planning of extracerebral tumors and assess its specific advantages and disadvantages. Materials and Methods: A MR system with an open configuration (OUTLOOK, Picker Int.) was installed in August 1996. The field strength is 0.23 T with a homogeneity of 5 ppm over 40 cm around the isocenter. The 5 G fringe field is 2.4 m x 2.3 m. The iron core electromagnet has a ramp time of 10 min. The room was supplemented with three sets of orthogonal lasers for patient positioning. An Ethernet link was installed to transfer imaging data in DICOM 3.0 format from the DEC Alpha 400 scanner computer system (Windows NT) either to the treatment planning workstations (Voxelplan, Helax TMS) or to a PC (Pentium 150) connected to the X-ray simulator. Software (IDL 4.01 for Windows NT) was programmed to measure and correct image distortion based on measurements of a specially designed 3D grid phantom and polynomial modeling. The corrected data sets were transformed from parallel projection into central ray projection after definition of the isocenter and the focus-isocenter distance. Patient placement in treatment position is achieved on a specially designed flat carbon table which fits onto the curved MR table. An acrylic head fixation device which allows the definition of discrete angulations of the C-spine was used for patients with head and neck tumors. Results: Image distortion calculated based on the phantom measurements increased notably with increasing distance from the isocenter of the magnet. It was below 1 mm within a sphere of 10 cm diameter and about 5 mm at 15 cm distance. Theoretical calculations were performed to calculate the error in dose application when treatment planning of thoracic tumors is performed solely based on the MRI information. The thoracic wall, lungs and mediastinum were segmented and homogeneous attenuation coefficients were assigned (1 for soft tissue, 0.25 for lungs). The error in dose prescription for standard sized treatment portals (> 8 x 8 cm2) as compared to CT based dose calculations was less than 3%. Conclusion: Open low-field MR systems are an attractive adjunct for radiotherapy treatment planning, not only for economic reasons. Besides the general advantages of MR imaging, i.e. soft tissue contrast and multiplanar imaging, they are comparatively inexpensive and can be easily integrated into existing facilities due to the small fringe field. The open geometry of the magnet allows unrestricted positioning of the patient and the use of standard positioning devices, which do not fit into a CT or standard MR scanner. Furthermore, the operator can interactively correct the placement of the patient. On the other hand, image distortion is more pronounced in this MR system as compared to conventional scanners due to the open configuration of the magnet. Another problem of the MR method as compared to CT is the lack of electron density matrix. Tissue inhomogeneities may introduce errors into the dose calculation process especially when small radiation portals are used. This may be less important for standard field sizes