$442 Posters

whether the improved accuracy of FDG-CT-PET for the localization of in situ rectal tumours does influence the extension of the GTV, the applied dose to the organs of risk and the decision making in the preoperative radiotherapy for locally advanced rectal cancer.

Conclusions: The integration of a combined FDG-CT-PET in the virtual simulation of preoperative 3D conformal radiotherapy of locally advanced rectal cancer may significantly reduce uncertainties in the GTV target delineation, thus improving accuracy of treatment volume definition. We will report on our first experience and the results of our pilot study on the basis of all patients treated in the time between March and October 2004.

1052 poster

Critical appraisal of CTV and PTV definitions by matching of simulator (SF) and portal films (PF) in prostate cancer patients (pts) M. Motta ~'2, S. Presilla 3, A. Franzetti 2, G. Ballerini 2, G. Pesce 2, A. Richetti 2, J. Bernier 2

~ Universita'degfi Studi di Milano Bicocca, U. O. Radioterapia, Monza, Italy (2) IOSI, Radio-Oncology, Bellinzona, Switzerland 310SI, Medical Physics Unit, Bellinzona, Switzerland

Background and objective: In prostate cancer radiotherapy accuracy and reproducibility of patient setup is of critical importance since these carcinomas require high doses of irradiation.Our purpose was to verify accuracy of CTV and PTV definitions through the evaluation of systematic and random errors occurring during treatment.

Materials: From September '03 till November '03, a retrospective analysis of PF taken every other day was performed in 18 pats with prostate cancer. Out of 350 SF and PF paired using EPID Markus I (Varian), 308 remained fully evaluable for matching procedures. Pts (Group 1 vs 2( were stratified as follows: PSA (< vs ->10 ng/mL), or Gleason Score (< vs -> 7), or T stage (T1-T2 vs T3). In Group 1 (n:3), pts received 76 Gy (38 frs/7.5 wks) to the prostate volume alone whilst, in Group 2 (n:13), 70 Gy (35 frs/7 wks) were delivered to prostate and seminal vesicles before a boost of 6 Gy/3frs to prostate volume alone.

Methods:

1) We calculated the mean deviation in the x,y axes and the mean rotation, also using absolute values, both at 0 ° and at 270 ° , in order to assess systematic and random set-up errors.

2) Applying a biquadratic model, we identified the sandard deviation (SD), absolute value, in the x, y and z axes. So we got the best treated and the worst treated patient, such as the one with the lowest and the greatest SD value respectively.

3) A new CTV (3SD-CTV) was then outlined in these 2 patients, expanding it isotropically by 3 times the standard deviation value.

Results:

1) Calculations of mean values gave the following patterns : At 0°: Mean deviation of the absolute value (A. V) of x:l .8 mm (range 0.2 + 3.7); Mean deviation of the A.V. of y: 1.3 mm (range: 0.2 +3.0); Mean deviation in x:1.1 mm (range:-2.4 + +3.8); Mean deviation in y:0.2 mm (range: -2.8 + +3); Mean rotation: 0.9 ° (range: -10 ° + 15°). At 270°: Mean deviation of the A.V.value of x: 1.0 mm (range 0.1 + 3,8). Mean deviation of the A.V. of y : 0.9 mm (range: 0.3 +3.0). Mean deviation in x:0.6 mm (range:-1.3 + +4); Mean deviation in y:0.1 mm (range: -1.6 + +3.0); Mean rotation: 0.9 ° (range: - 15 ° + +13),

2) The lowest and the highest SD were 1.4 and 4.3 mm respectively. The best treated patient belonged to group 1, while the worst treated one to group 2.

3) The 3SD-CTV was inside the PLANNED PTV for the best treated patient, and inside the 2SD CTV (comprehensive of the 95% of the population) for the worst treated one.

Conclusion: This study on systematic and random setup errors shows the relevance of an adequate selection of CTV and PTV delineation criteria in prostate cancer patients.

1053 poster

Organ contouring for prostate cancer: interobserver and internal organ motion variability

M. V. Iba~ez-Guaf , X. GuaI-Arnad, F. Lliso 2, S. Rold~n 2

~ University Jaume I, Mathematics, CastellSn, Spain 2Hospital La re, Servicio de Oncologia Radioterbpica, Valencia, Spain The interobserver variability inherent to manual delineation of normal tissue anatomic structures, has been shown to influence in the treatment process for prostate cancer. Moreover, prostate motion itself during the course of treatment may lead to variations in margin determination.

Our data consists on CT parallel sections of a human prostate and bladder. For determining the interobserver variability, the contours of the organs in all sections have been manually delineated by 4 independent observers. For deriving the variability due to organ movement, CT sections have been acquired on a same patient during 7 different sessions.

To derive margins around the prostate and bladder which include the above geometrical uncertainties, the following mathematical methods have been used. The radius-vector functions have been used to define a mean contour; the standard deviation from all the radial directions is computed. Since, in the medical practice just the variation in 4 directions is considered, the standard deviation function is fitted by an ellipse. Finally the ellipse is placed on the prostate or bladder contour to obtain the desired margins. An unbiased stereological estimator for the volume of the organs is also computed, which will allow us to obtain an error approximation of the estimator, depending on the CT slice thickness and the contours delineation.

Although the topic of this work has been widely considered in the literature, our method has in our opinion two main advantages. The first one is that we compute variability in all directions and the second one is that all this information is used to obtain the margins following the techniques and limitations imposed by the medical practice.

The main conclusions are that the margins arising from the organ motions in this pathology are significantly greater than those coming from interobserver variability, and that the tools developed in this work allow to obtain in a efficient way the require d margins from a set of CT slices.

1054 poster

Impact of MRI/CT co-registration on target volume definition and radiotherapy planning of tumours located close to the base of skull M.T. Guerrero Urbano 1, C. Kong 2, K. Burke 2, C. Nalder 2, S.

1 2 1 Reinsberg , L. Moore, H. McNair , A. Sohaib 2, K.J. Harringtod, C.M. Nutting 2

~ lnstitute of Cancer Research/Royal Marsden hospit, Radiotherapy, London, United Kingdom