Abstract book - ESTRO meets Asia

S43 ESTRO meets Asia 2018

approaches could give better results. Thus, we investigated different approaches to determine the maximum performance of the ABS software. The study aim was to evaluate the contouring accuracy of ABS software by using two different approaches for a treatment planning system. Material and Methods Volumetric computed tomography datasets for 30 patients with prostate cancer treated with intensity-modulated radiation therapy were retrospectively reviewed. Seven anatomical structures in the prostate region were contoured as structures: the prostate, seminal vesicle, rectum, bladder, pubic, ischium, and femoral head. All structure sets of 30 patients were manually contoured by radiation oncologists. The structure sets of 20 patients randomly selected were used in RayStation version 4.7.4.4 (RaySearch Laboratories, Stockholm, Sweden) for two reference atlases: group (a), 20 reference structure sets from each of the 20 patients; and group (b), 160 reference structures (8 structures × 20 patients) were created by adding each structure. Two different ABS approaches were performed for 10 patients excluding 20 patients who added reference atlases: batch ABS (B-ABS) in which the most similar metric from group (a) was used to select the structure set, and multi-segment ABS (Ms-ABS) in which the most similar metric from group (b) was used to select the structures. For the evaluation of ABS accuracy, similarity was assessed by using the Dice similarity coefficient (DSC). The Wilcoxon signed rank test was used to compare the difference in the DSC values between the B-ABS and Ms-ABS approaches. Two-sided P-values <0.05 were considered as indicative of statistical significance. Results In the B-ABS approach, the DSC values indicated substantial agreement (>0.7) for the pubic, ischium, and femoral head; moderate agreement (0.7–0.5) for the prostate, rectum, and bladder; and large variation (<0.5) for the seminal vesicle. In the Ms-ABS approach, the DSC values indicated substantial agreement for the prostate, rectum, pubic, ischium, and femoral head and moderate agreement for the seminal vesicle and bladder. There were significant differences in the DSC values between the B-ABS and Ms-ABS approaches: prostate, 0.79 vs. 0.58, P < 0.01; seminal vesicle, 0.52 vs. 0.24, P = 0.02; rectum, 0.76 vs. 0.54, P < 0.01, respectively. Conclusion The Ms-ABS approach improved the contouring accuracy of the target and organs at risk for prostate cancer. OC-113 Implementation of forward planned IMRT to treat breast and chest wall patients. L. Andrews 1 , A. Williams 2 , M. Satterthwaite 1 , J. Donaldson 2 , N. Whitaker 1 , L. Seddon 1 , D. Paterson 1 , C. Cowles 1 1 Wellington Regional Hospital- Blood and Cancer Centre, Radiation Therapy, Wellington, New Zealand 2 Wellington Regional Hospital- Blood and Cancer Centre, Medical Physics, Wellington, New Zealand Purpose or Objective To describe the implementation of forward planned IMRT (FIMRT) for breast and chest wall treatments. Historically, 3DCRT tangential fields using a combination of energies, EDW, MLC and a fixed normalisation point have been used to treat the breast and chest wall. FIMRT planning was investigated for its potential to improve target coverage, planning and treatment efficiency and reduce the use of 18MV beams. Material and Methods 1) Definition of a volume of Interest (VOI) : Planning target volumes (PTV) are not routinely delineated for breast treatments at our institution. Therefore, a method of creating a field defined breast VOI was devised based on the 50% isodose. This VOI was compared to PTVs drawn

by our speciality breast Radiation Oncologists (RO). 2) Comparison of FIMRT and irregular surface compensator (ISC) : Two IMRT techniques were investigated and compared to 3DCRT. FIMRT and ISC plans were retrospectively created by multiple staff (radiation therapists and physicists) on 13 datasets. Plans were prescribed to D 50 as per ICRU 83. The planning aims were to cover the VOI with between 95% and 107% of the prescribed dose. Plan quality was assessed using the VOI D 98% , D 2cc and D 50% . Plan efficiency was assessed using the time to plan, time to treat and the number of MU per field. 3) The impact of breast swelling was investigated: Once the preferred technique was selected, breast swelling of up to 1.3cm during treatment was simulated in the planning system by modifying the patient contour, and its effect on the VOI dose distribution recorded. 4) Method to allow larger patients to be treated with IMRT developed : For the patient cohort that had large separations and adequate 95% coverage could not be achieved with 6MV FIMRT alone, a technique using a mix of 6MV FIMRT and 18MV open fields was developed. Results 1) The breast VOI was comparable to manually contoured PTV volumes. The VOI provided adequate delineation and saved considerable planning time for the RO. The RO is able to edit this volume. 2) Table 1 and 2 summarises the comparison between 3DCRT and IMRT plans. Both FIMRT and ISC techniques could be used to create good quality plans with 6MV only for patients with a separation up to 24cm. However, the FIMRT technique was quicker to plan and needed less MU for equivalent dose distributions. In addition, FIMRT was preferred by planners as it utilised the planners' knowledge of conventional planning to drive the system to obtain the desired result in an intuitive way.

3) Breast tissue swelling as simulated was found to not adversely affect the planned dose distribution. 4) The addition of 18MV open fields allowed patients with separations of up to 28cm to be treated. The amount of 18MV needed to be increased with separation, from 0% at 24cm up to 50% at 28cm. Conclusion The project has resulted in the successful implementation of FIMRT planning for all breast and chest wall patients except those with separations larger than 28cm. OC-114 Blended learning in radiation therapy education B. Mudie 1 1 University of Otago, Radiation Therapy, Wellington, New Zealand