Abstract Book

S1198

ESTRO 37

that the deviation of detectability is the smallest and the deviation of the severity is largest, so that the value itself is due to the order of S> O> D. In the present study, the stage of rectal cancer treatment is based on Gangdong Kyunghee University Hospital, and further studies are needed to standardize. EP-2169 Data quality of DICOM RT data bank and manually reported values – results from the DBCG HYPO trial C. Brink 1 , E.L. Lorenzen 2 , S.L. Krogh 2 , J. Westberg 2 , M. Berg 3 , I. Jensen 4 , M.S. Thomsen 5 , E.S. Yates 5 , B.V. Offersen 6 1 University of Southern Denmark, Institute of Clinical Research, Odense, Denmark 2 Odense University Hospital, Laboratory of Radiation Physics, Odense, Denmark 3 Vejle Hospital, Department of Medical Physics, Vejle, Denmark 4 Aalborg University Hospital, Department of Medical Physics, Aalborg, Denmark 5 Aarhus University Hospital, Department of Medical Physics, Aarhus, Denmark 6 Aarhus University, Institute of Clinical Medicine, Aarhus, Denmark Purpose or Objective Radiotherapy dose response models tend to rely on the entire dose distribution and not just few dose parameters. Currently, national DICOM RT banks are emerging to support access to the entire dose distribution. Data quality is obviously important within research, thus the data quality of the data banks should be evaluated. The current study reports the data quality of manually reported data versus data submitted to a national radiotherapy data bank. Material and Methods The current study is based on data from the 1522 Danish breast cancer patients, included in the DBCG HYPO trial. For all patients included in the trial specific dose parameters were prospectively reported manually. During the trial it was decided to utilize the national data bank for the Danish patients in the trial making it possible to compare the two data sets. In addition to dose parameters, laterality of the treatment (left/right), treatment arm (40/50 Gy), and whether a boost had been prescribed was extracted from the DICOM data in the data bank. Results Of all the categorical values (laterality, trial arm, and boost) there were only 25 inconsistencies between the manual and data bank values. The data bank contained the correct value for 23 of these inconsistencies. For the two remaining cases, the data bank stated incorrectly that a boost was not delivered. The two boosts were manually calculated electron boosts, and without a boost volume delineated in the planning CT, thus boost delivery data was not in the data bank. For the continuous data, the fraction of outliers and extreme outliers between manually reported and data bank values were calculated. These were defined as data differences outside box-plot- whiskers of 1.5 or 3 times the interquartile range, respectively. Except for date of RT and volume of CTV which had ~5% and ~2% of outliers and extreme outliers, the typical fractions were ~20% and ~10%, respectively. A major part of the differences could be related to either mistyping of numbers (e.g. interchange of digits), while others were related to a misunderstanding of which value to report (e.g. V95 or 1-V95). However, some of the deviation was also related to uncertainty of reported max dose values, which are sensitive to the specific algorithm used to calculate these values. Checks on all dose parameters showed consistency between treatment plans used for manual reporting and those in the data bank.

Conclusion The data quality of the data bank is superior to the quality of the manually reported values - in particular for the continuous data. In the current study, 10% of the continuous data were classified as extreme outliers and could influence results from clinical trials. However, some of the wrong manual data would have been detected, since the values were off by orders of magnitude, but the majority would be hard to detect without input from the data bank. The superior data quality of the data bank, and the detailed information available, support the use of data banks for future research. EP-2170 A feasibility study of optical surface imaging technology in radiotherapy L. Yang 1 , G. Han 1 , Q. Song 1 , C. Ruan 1 , L. Li 2 , R. Zhong 2 , S. Bai 2 1 Renmin Hospital of Wuhan University, Department of Oncology, Wuhan, China 2 West China Hospital of Sichuan University, Radiation physics technology center, Chengdu, China Purpose or Objective CBCT is a standard and effective method in IGRT, however, it is time-consuming and causes an additional dose, meanwhile, it cannot be monitored the errors. Catalyst™ system is a novel optical surface imaging system for patient positioning and surveillance during radiotherapy. It has some attractive features in radiotherapy for patient positioning such as set-up, monitoring and gating, however, the application of it is also limited by relative movement of the skin and the deep tumor. In our study, we compare set-up displacements derived from Catalyst™ system to displacements derived from CBCT in order to explore the