ESTRO 37 Abstract book

ESTRO 37

S597

PO-1063 A “big-data” platform, managing the clinical data & workflows and facilitating clinical research L. Persoon 1 , H. Kooy 2 , F. Van der Kruijssen 1 , J.W. Doosje 1 , Wolfgang 2 1 ICT Group N.V., Healthcare, Eindhoven, The Netherlands 2 Massachusetts General Hospital, Department of Radiation Oncology, Boston, USA Purpose or Objective Today’s radiotherapy is a complex, data driven clinical endeavor. The escalation of innovative technology in radiotherapy has resulted in a large, diverse array of different computer systems used in concert for a single point of care. Interoperability between the different systems and the intercommunication between them has been identified as one of the key-factors to safer and more efficient treatment and where failed communication has been shown to account for the majority of radiotherapy errors (38%). Furthermore, intercommunication and data management loom as two of the biggest impediments in clinical implementation of adaptive radiotherapy (ART). The purpose of this study is to show that management of the clinical workflow and the produced data using novel and standard data- management paradigms creates a safer and more efficient clinical environment and simultaneously enables a big-data analysis platform to improve personalized medicine. Material and Methods In this study, a radiation oncology collaboration platform (ROCP) was developed using InterSystems technology (InterSystems, Cambridge, MA, USA) and Google Angular2 (Google, Mountain View, CA, USA). The ROCP implements a novel and comprehensive data management approach by implementing the DICOM RT 2 nd generation RT course model. Clinical workflows can dynamically be modelled to support advanced clinical procedures such as ART, utilizing a radiotherapy specific domain design and language. 3 rd party vendor applications are integrated into the Service Oriented Architecture of the ROCP, using interoperability standards (e.g. DICOM Unified Procedure Steps(UPS)) to convey the task and necessary data. Departmental efficiency with the ROCP was evaluated by collection of task completion time as well as the number of delayed treatments. Results The proposed solution tested with a proof-of-concept system in a pilot showed that task compliance in key areas, such as contour definition, radiation dose prescription, IMRT QA measurement completion and pre- treatment plan checking rises to levels above 95%. In parallel, the percentage of delayed treatments dropped significantly from 13% to 6% for IMRT plans and from 10% to 1% for 3D-CRT plans. An absolute risk of delay reduction of 7% for IMRT plans and 9% for 3D-CRT plans and a relative risk of delay of 0.46 for IMRT plans and 0.10 for 3D-CRT plans was observed. Conclusion Managing the clinical workflow and the produced data shows a drastic reduction in treatment delays, gaining workflow efficiency. Besides the gain in efficiency, collection of data via workflow management introduces a well formed patient clinical model allowing for the analysis and detection of errors, improved efficiency necessary for routine ART, and establishing a valid, clinical data model for personalized medicine using Big Data analysis as Big Data begins with management of patient data within a well-defined clinical context.

Poster: RTT track: Imaging acquisition and registration, OAR and target definition

PO-1064 The utilization of preoperative MRI in breast lumpectomy cavity contours after conserving surgery W. Huang 1 , Y. Dong 2 1 Shandong Cancer Hospital, Radiation Oncology 6, Jinan, China 2 University of Jinan-Shandong Academy of Medical Sciences, School of Medicine and Life Sciences, Jinan, China Purpose or Objective To compare the lumpectomy cavity (LC) and planning target volume (PTV) defined by preoperative magnetic resonance imaging (pre-MRI) and postoperative prone computed tomography (post-CT), in order to optimize the tumor bed delineation for radiotherapy planning after breast-conserving surgery. Material and Methods 16 patients were included with early-stage breast invasive ductal cancer, who have undergone breast- conservative surgery with titanium clips placed around the tumor bed, all of them received the pre-MRI and post-CT in prone position. The CT and MRI sequences (T1, T2, T2W-SPAIR, DWI, DCE, sdyn-eTHRIVE- named by Philips- Fast Dynamic Contrast-enhanced Subtraction MRI Imaging) were manually delineated respectively. Various PTVs were defined as unified margins of 15 mm expanded based on the corresponding LCs. Differences were measured in terms of consistence index (CI), dice coefficient (DC), geographical miss index (GMI) and normal tissue index (NTI).

Results The CIs for T1, T2, T2W-SPAIR, DWI, DCE, sdyn-eTHRIVE and CT were 0.646, 0.648, 0.680, 0.676, 0.718, 0.716, respectively, implying that the DCE and sdyn-eTHRIVE could specifically show the preoperative details of the mass, closely followed by DWI. The DCs between the volumes defined by CT and the two MRI sequences (DCE and sdyn-eTHRIVE) were 0.67± 0.20, 0.66 ± 0.22 for LCs and 0.87 ± 0.06, 0.86 ± 0.07 for PTVs, respectively. The GMI and NTI based on T1 and T2 were more obvious than the other MRI sequences with CT. Conclusion The pre-MRI does improve the target volume delineation, especially for that large, pendulous, breast size. The target volumes of LC and PTV generated according to a MRI sequence are smaller than those according to CT mostly. The DCE and sdyn-eTHRIVE both might be the better sequence for precise definition of LC. The PTV-MRI volumes, defined by the DCE and sdyn-eTHRIVE, were