1. MRI SIMULATION FOR CONFORMAL RADIATION THERAPY OF PROSTATE CANCER Pasquier D 1-3, Palos G 3, Castelain B 1, Lartigau E 1,2, Rousseau J 2,3 1 Department of Radiotherapy, Centre O. Lambret, 2 University Lille II, 3 ERT 23 0.6 0.3 2.9 0.8 1 1.6 0.6 0.4 3.5  y (mm)  x (mm) 0.6  y (mm)  x (mm) 63 613 54063 613 53063 613 56063 613 500 Resonance frequency (Hz) 1.8 0.2 Bladder 2.41.22.51.9 Outer contour Volunteer 4Volunteer 3 Volunteer 2Volunteer 1 Purpose/Objective Thanks to its excellent soft tissue contrast, MRI ensures better delineation of target volumes in many locations, such as the prostate. MRI is only used, in most cases, in conjunction with Computerized Tomography (CT). MRI simulation would eliminate the localization errors introduced by image matching and also makes it possible to dispense with an additional imaging examination that is costly, time consuming, and which generates additional irradiation. The obstacles mentioned are geometrical distortion and chemical shift, measurement of electron densities, and the compatibility of some treatment planning systems (TPS). Materials/Methods Geometrical distortion measurements were carried out on two 1.5 T MR scanners, a Magnetom Vision Siemens ® and a Gyroscan Intera Philips ®. The phantom used, measuring 400 mm x 300 mm x 210 mm, was composed of 730 glass spheres immersed in a 1,2-propandiol solution. The phantom was imaged with a body coil with a T2-weighted sequence conventionally used for prostate imaging (Turbo Spin Echo, 4mm thick contiguous axial, 512x512 matrix, Field Of View 45 cm). The effect of chemical shift and magnetic susceptibility were evaluated in four healthy volunteers on the Siemens ® 1.5 T Magnetom Vision scanner. Use was made of T2 weighted TSE sequences (FOV 40x40 cm, 5mm thick contiguous axial slices, 256x256 or 512x512 matrices). By swapping over the respective phase and frequency encode gradients, the direction of chemical shift and magnetic susceptibility were changed, which it was thus possible to quantify for prostate, bladder and outer body contour. To assess the effect on dosimetric calculations of uncertainty in determining electron densities, the method used was therefore to assign, in the CT images, different relative electron densities for the soft tissues (SED) and for bone (BED). Treatment plans with identical weightings were then calculated and compared with the treatment plan devised with the initial CT. Dosimetric calculations were carried out on a Helax ® 6.0 TPS. As this system does not accept MR images, DICOM headers of the MRI files were converted to make them similar to those of CT images, while preserving the geometrical characteristics of the original images. Results Geometrical distortion remained at small values: less than 2 mm and 3 mm for FOV of 20 cm and 45 cm (Fig 1). The chemical shift and magnetic susceptibility values obtained, ranging from 0.3 to 3 mm for outer contour, were well below the theoretical values: MR machine tunes the excitation frequency to the subject so as to obtain maximum radio frequency reception, minimizing chemical shift (maximum difference = 60 Hz between volunteers 1 and 2) (Tab 1). The chemical shift of prostate was too low to be measured. By assigning a relative electron density of 1.02 to all the tissues, i.e. without taking tissue heterogeneities into account, an underdosage of the target volume of 2.8% was obtained. By assigning to the soft tissues and to the pelvic bones the values recommended by the ICRU, the dosimetric deviation was lowered to 0.3 %. 10% overestimation or underestimation of the relative bone densities leaded to overdosing and underdosing by 1.7 % and 0.8 %, respectively. The same overestimation or underestimation for the soft tissues leaded to dosimetric deviations of about 2%. Variation of 25 % of bone or soft tissues relative densities leaded to dosimetric deviations about 3 to 5 %. The assignment of ICRU relative electron densities to only two structures in MR images seems to permit dose planning that is identical with that obtained with CT (Fig 2). Similarly, DRR can be produced after the assignment of relative electron densities (Fig 3). Conclusions None of the technical obstacles mentioned (distortion linked with the system, chemical shift and magnetic susceptibility, lack of information on electron densities) represents a stumbling block. MRI alone should be used in a near future in planning and virtual simulation of prostate cancer. Figure 1 Table 1 Figure 3 Figure 2