ESTRO 37 Abstract book

S1121

ESTRO 37

EP-2046 Robust optimization (4D) IMPT strategies using full and empty bladder CT scans for cervical cancer C.L. Brouwer 1 , W.P. Matysiak 1 , P. De Boer 2 , J.A. Langendijk 1 , J.C. Beukema 1 , S. Both 1 1 University of Groningen- University Medical Center Groningen, Department of Radiation Oncology, Groningen, The Netherlands 2 Academic Medical Center, Department of Radiation Oncology, Amsterdam, The Netherlands Purpose or Objective External beam radiotherapy of cervical cancer is challenged by large inter- and intra-fraction motion, mainly due to differences in bladder and rectal filling. To account for target volume shape and position variations an appropriate IGRT strategy has to be applied. In proton therapy the corrections for the resulting changes in dose conformality are typically more critical than in photon therapy. This in silico planning study was performed to evaluate different robust optimization (4D) IMPT strategies using full and empty bladder CT scans for locally advanced cervical cancer patients in terms of target coverage and OAR dose. Material and Methods CT scans of five patients were retrospectively selected for this study. The ITV 45 was defined as GTV cervix + 1cm with the addition of uterus, upper vagina, and lymph nodes. IMPT plans were created using two lateral and two posterior oblique fields in RayStation version 5.0 with 5 mm setup and ±3% range robust optimization parameters. In addition, the 4D optimization option available in RayStation was used to create plans which were optimized on two CT sets simultaneously: the plan CT and a selected repeat CT having typically a different bladder volume. Six types of plans were created in total: a full bladder single-CT-optimized IMPT plan (F), a full bladder 4D optimized plan where the repeat CT was chosen such that the bladder volume difference between the two sets was less than 100 cc (F100), and the same type of plan but with the bladder volume difference in the interval of 100-200cc (F200). A corresponding set of plans was also created with the empty bladder planning CT: (E), (E100), and (E200), respectively. Each plan type was evaluated on each of the repeated CT scans and the target coverage of the ITV 45 as well as OAR doses (whole bowel, bone marrow, rectum, bladder) were computed. Plans were optimized to cover 98% of the ITV with 95% of the dose (voxelwise min evaluation), while an effort was made to minimize the whole bowel and bone marrow dose. Results The 4D optimized plans resulted in better target coverage than the non-4D optimized plans (see Figure 1A). Overall, V95 of the ITV45 was highest for the F200 and E200 plans (98.1 and 98.3, respectively). The target coverage was adequate for all planning strategies as long as the bladder volume difference between the plan CT and the evaluation CT was maintained within 100cc. Empty bladder plans resulted in slightly higher whole bowel dose while the bone marrow dose was slightly lower (Figure 1B) , as a result of decrease in the radiological equivalent pathway.

breast positioning. To the best of our knowledge, morphological variations of the breast and their impact on the treatment quality remain unknown. We propose here an original method based on optical 3D surface scanning to get objective measures of the irradiated breast volume evolution and its local deformations. Material and Methods We developed a software using the dosimetric CT structures and patient surface scans ( S3D ) acquired during the treatment. The hand-held, optical, non ionizing surface scanner Artec (R) Eva was chosen for its quick acquisition and infra millimeter precision and resolution characteristics. When realizing the dosimetric CT, a metal wire was placed on the patient’s skin to circumscribe the palpated gland. Surfaces were aligned with manual and automated registration. Then a surface correspondence algorithm was applied. The wire delimiting the targeted gland on the dosimetric CT was then virtually identified on a S3D selected as reference ( refS3D ) thanks to correspondence data from CT to S3Dref , then identified on other S3Ds (correspondence data from S3Dref to each S3D ) to recognize breast position and perform volume evaluation. Deformations were evaluated by 325 measurements (each 10 degrees on a half-sphere) from the center of the breast base to the skin. To evaluate our software we performed a dosimetric CT, contoured the skin, the targeted breast and wire, then performed a S3D . For all tests, the S3Dre f was exactly the same. Volume and deformation measurement reproducibility was tested on 5 strictly identical S3Ds (and identical to the S3Dref ). Robustness to extra breast misposition was tested on 5 S3Ds progressively deformed. Then, the ability to measure targeted breast morphology was tested on 5 S3Ds with known volume variation. Results Reproducibility tests - We found a global mean volume deviation of +2.9% ([1.2% ; 3.3%], σ = 0.82). A mean error of +0.04mm (σ = 3.71) was observed in deformation measurements. Robustness to mispositions - We described a mean volume deviation of -1.8% ([- 11.1% ; 4.2%], σ = 5.37). A mean error of -0.36 mm (σ = 3.01) was observed in deformation measurements. Ability to follow breast volume changes - The software had a mean volume deviation of +3.37% ([0.04% ; 5.84%], σ = 2.25).

Figure: Volume variations measurement

Conclusion We propose a reproducible and clinically relevant method to follow irradiated breast volume and morphology during radiation therapy after con-servative surgery for breast cancer. By the use of non ionizing surface scanner, we could propose, in the future, a safe and easy to implement clinical trial to study breast deformations and volume changes and their impact on the treatment quality.