Proceedings of the 40th Annual ASTRO Meeting 2131 A PROSPECTIVE COMPARISON OF 3 SYSTEMS OF PATIENT IMMOBILIZATION FOR PROSTATE RADIOTHERAPY 293 S. Malone, J. Szanto, L. Gerig, G. Perry, S. Manion, J. Crook Ottawa Regional Cancer Centre, Ottawa, Ontario. Purpose/Objectives: To compare prospectively 3 systems of patient immobilization for conformal prostate radiotherapy. Materials & Methods: 80 patients with prostate cancer, treated with radical radiotherapy, were entered into this 3 arm study of patient immobilization. In two methods of immobilization, foam robber leg-support and full body alpha cradle, the patients were treated supine. In the third method, Thermoplast Hip-Fix, patients were treated prone. This study examined treatment setup error for each of the three arms, where setup error is defined as the geometric difference between simulation and treatment field position with respect to bony landmarks. The setup error was determined from twice weekly port films, analyzed with NIH image software. Error is reported as the total vector error as well as the individual Cartesian components (x=medio-lateral, y=anterior-posterior, z=superior-inferior). Daily Setup Errors Total Vector Error >5mm >10mm >5ram >10mm X Y Z X Y Z Leg Cushion (20Pts) 16% 18% 20% 3% 2% 4% 60% 16% Alpha Cradle (25 lots) 7% 12% 17% 1% 4% 1% 54% 8% Hip Fix (35 Pts) 2% 3% 4% 0% 0% 0% 35% 0% Summary: The 2 methods of rigid patient immobilization (Hip-Fix and alpha cradle) tested in this study show a reduction in setup error over what is reported in the literature for free set-up. In our experience Hip Fix thermoplast immobilization was the best system of patient immobilization with a significant reduction in positioning error. This study lends strong support to the utility of rigid immobilization systems for pelvic radiotherapy. 2132 PORTAL IMAGING FOR EVALUATION OF DAILY ON-LINE SET UP ERRORS AND OFF-LINE ORGAN MOTION FOR RADIOTHERAPY OF CARCINOMA OF PROSTATE. H. ALASTI, *C, CATTON, *,*N. MIDDLEMISS,**G. o'VrEWELL, **L. GREEN, *P. WARDE DEPT. OF CLINICAL PHYSICS,*DEPT. OF RADIATION ONCOLOGY, **DEPT. OF RADIATION THERAPY,THE PRINCESS MARGARET HOSPITAL,TORONTO,ONTARIO, CANADA PurpoSe: This study is designed to measure the daily on-line treatment set up uncertainty and off-line prostatic motion for 50 patients, using portal images acquired over the entire course of 42 fractions of escalated dose conformal radiotherapy. On the basis of this large patient database, we determine an optimum margin for Planning Target Volume (PTV) that incorporates both set up error and organ motion. Methods and Materials: The patient is immobilized with an alpha cradle and three fiducial markers are inserted in apex, base and mid- prostate. The treatment plan is a coplanar isocentric 6-field technique on 18 MV photons, using a multi leaf collimator to conform to the shap~ of the field. For better delineation of the Gross Target Volume (GTV), we have optimized the CT scanner parameters (mA, slice thickness and reconstruction filter) to improve both the low contrast and spatial resolution of the image. We used the portal imager from Varian and because of inherent low contrast and poor visualisation of the portal imager, we enhanced the image quality by modifying and optimizing the image acquisition parameter set up of detectors and by adopting an individual calibration mode for each lateral. For further improvement of image quality we treat and image a lateral on 6 MV photons, alternating between right (RT) and left (LT) lateral. Two lateral portal images are captured daily, one on 6 MV the other on 18 MV. The image quality for 6 MV is good and is used for image matching. For set up error analysis, the RT and LT simulator radiographs are digitized and downloaded to the portal imager as a reference image for matching with the on-line portal image. Bony landmarks selected on the reference image are aligned and superimposed on the anatomical landmark of the corresponding on-line portal image. Set up errors from mismatch relative to bony structures are calculated. The seeds are not reliably observed in the on-line image due to their small dimensions (.8mm dia. and 3mm length) and the detector low resolution. Therefore the prostatic motion will be evaluated by processing the stored portal images retrospectively. Results: We have treated 20 patients and acquired total of 1600 lateral portal images. The image quality captured on 6 MV is good and used for daily matching. The systematic error for Anterior-Posterior (AP) and Superior- Inferior (SI) ranges from .1 to .5 mm, systematic errors of about tmm occurred for two patients. The maximum measured random error on AP direction is 6 mm and on SI direction is 5 mm. The systematic error for the AP rotation is between .1 ° to .4 °. Maximum measured AP rotation is 6 °. In 80% of the total measurements, the displacement for AP and SI is _< 2 mm, only 2% of measurements on AP and 1% on SI has errors _> 5 mm. The off-line organ motion will be assessed by processing and enhancing the stored portal images for fiducial marker detection. Conclusion: We have established a portal imaging protocol and trained our radiation therapists to accept or adjust a treatment set up. The study shows that on-line daily imaging has reduced the systematic error significantly. The typical range of displacements along AP or Sl is about 2 mm and for the AP rotation is 2 °, indicating that careful daily positioning and using alpha-cradle immobilizer was effective. In addition to off-line analysis of organ motion, we will be evaluating the new generation of portal imager from Varian for on-line visible detection of the fiducial markers. Furthermore, we will increase the seed size to improve detection. From the combined set up error and organ motion measurements in a patient population we will determine an optimum PTV margin for the prostate and the technique used. and assess the adequacy of the currently used margins.