S 174 IGRT: TREATMENT DELIVERY AND PATIENT POSITIONING

into our in-house developed image-guided workstation. Regions of interestwere placed around the target and each OAR. Local rigid CT to CBCT reg-istration was performed to quantify the treatment position of each OAR aftercorrectly positioning the target. Delineated structures of OARs were trans-lated accordingly. DVHs were recomputed assuming shift invariance of thedose distribution and checked against treatment planning constraints. Finally,the shift invariance assumption was validated by recomputing the dose.Results: OAR positions during planning (thin lines) and during treatment(thick lines, shifted based on CT to CBCT registrations) overlaid on the midventilation planning CT, are shown in Fig 1. The situation shown here wasnot clinically accepted as residual OAR misalignment exceeded the limit of10 mm in the AP direction. DVH evaluation, however, showed that OAR doseconstraints were not violated. The maximum esophagus fraction dose was5.0 Gy at planning and 4.6 Gy during treatment. The maximum spinal cordfraction dose was 4.0 at planning and 4.6 Gy during treatment. Both are wellwithin the clinical constraints of 9 Gy for esophagus and 6 Gy for spinal cord.Also in the other evaluated patients, geometrical limits proved too conserva-tive. There was a reasonable shift-invariance of the OAR dose (< 8%) for allfractions but one, where the dose was overestimated by 35% compared toPinnacle.

Conclusions: We developed an online dose-guided correction protocol thatprovides a realistic evaluation of the safety of residual OAR misalignmentsand thus simplifies the decision process.Evaluation showed that clinically ap-plied geometrical limits are often too conservative.

468 poster (Physics Track)

OPTIMAL ISOCENTER POSITIONING USING A FULLY AUTOMATEDPROCEDURE FOR LASER CENTER TO ISOCENTER SHIFTSK. Pasma1, S. Postma2, M. Luesink1, R. Veenstra1

1 INSTITUTE FOR RADIATION ONCOLOGY ARNHEM, Arnhem, Netherlands2 CABLON MEDICAL, Leusden, Netherlands

Purpose: Normally a CT simulator is used to define and mark the laser centeron the patient. During treatment planning the isocenter is set equal to thislaser center. If that isocenter position is not optimal (1) asymmetric fields areused or (2) a new isocenter is chosen and marked on the patient using anextra conventional simulator procedure. We implemented a new automaticprocedure which allows the treatment isocenter to be chosen independent ofthe laser center. The shift is applied using computer controlled motion of thetreatment couch. The procedure facilitates the use of standard anatomicalmark points on the CT simulator and optimal positioning of the isocenter afterthe Clinical Target Volume (CTV) and Organs at Risk (OAR’s) are delineated.Materials: The treatment isocenter is selected in the treatment planning sys-tem (Pinnacle 8.0m, Philips Medical, Eindhoven, The Netherlands). The vec-tor between the CT laser center and this isocenter is sent to the EPID com-puter using a script (G. Meijer, Catharina Hospital, Eindhoven). No manualinput is required. Furthermore, Digital Reconstructed Radiographs (DRR’s)are sent to the EPID computer. The treatment units are equipped with a novelsystem (TCSA, Cablon Medical, see screenshot) which allows automatedcouch movements controlled by the EPID workstation with an accuracy of0.3 mm. At the treatment unit, the patient is aligned on the skin marks us-ing the laser lines. TCSA then performs the shift from the CT laser center to

the treatment isocenter. For the first 2-3 fractions EPID images are acquiredand matched with the corresponding DRR’s to verify the couch shift (attentionlevel 6 mm for patients with mask fixation and 8 mm for others) and for an offline setup correction protocol. The absolute and relative couch positions arelogged in the EPID database for each fraction. The shift derived from theoff line setup correction protocol is incorporated from fraction 3 (patients withmask fixation) or 4 (others) onward.

Results: The procedure has been applied clinically for more than 100 pa-tients. No errors have been detected. The time required to shift the couchranges from 8 to about 22 seconds (~2-14 cm).Conclusions: This novel procedure allows the technicians to choose theoptimal treatment isocenter without compromise after delineation of the CTVand OAR’s. Asymmetric fields or an extra conventional simulation session areavoided. Standard anatomical mark points for each treatment site can be setat the CT simulator allowing a faster and more accurate marking of the CTlaser center. The extra workload on the treatment unit is minimal. No manualinput is required. The safety of the procedure is clinically proven.

469 poster (Physics Track)

PRONE AND PRONE-LATERAL PATIENT POSITIONING FORWHOLE BREAST IRRADIATION (WBI): A STEEP LEARNING CURVEL. Veldeman1, B. Speleers1, M. Bakker1, A. Impens1, S. Nechelput1, C. DeWagter1, R. Van den Broecke2, G. Villeirs3, W. De Neve1

1 UZ GENT, Department of Radiation Oncology, Gent, Belgium2 UZ GENT, Department of Gynaecology, Gent, Belgium3 UZ GENT, Department of Radiology, Gent, Belgium

Purpose: The objective of this study was to improve prone set-up for WBIand to asses set-up precision.Materials: Patients with early-stage breast carcinoma presenting for WBI af-ter tumorectomy without the need for irradiation of the lymph node regionswere included. All patients were planned in prone and supine position, butthe aim was to treat patients in prone position. The HorizonTM breast board(CivcoTM) was used for prone positioning. A standard positioning procedurewith laser marks on the thorax and breast of the patient was used for set-upin the treatment room. Set-up precision in prone position was verified by dailycone-beam CT. During the study we encountered some problems concern-ing the position of the heterolateral breast, set-up precision and body rotation(roll).Results: The position of the contralateral breast was improved by develop-ing a unilateral breast holder in cooperation with the company Van de Velde(Schellebelle, Belgium). Daily cone-beam CT showed large systematic andrandom set-up errors with a standard positioning procedure with laser markson the thorax and breast. We evolved to a direct breast positioning procedurewith only marks on the treated breast. Results showed less systematic andrandom set-up errors. Roll to a prone-lateral instead of a pure prone positionresults in the treated breast being shifted laterally and the untreated breastbeing elevated, allowing the use of horizontal beam directions that do notcross the treatment table. To obtain this prone-lateral position, modificationsto the breast board had to be made. Dose-indices for heart and lung showedsignificantly better results in prone position.Conclusions: The heart and lung could better be spared in prone position. Aunilateral breast holder improved the position of the untreated breast. A directbreast positioning procedure with only laser marks on the breast resulted inbetter set-up precision. A prone-lateral position rather than a pure proneposition was preferred.