Click here to


Are you sure ?

Yes, do it No, cancel

High Resolution Spectroscopic MRI and Scripting: Towards Automated Planning for a Multi-Center Clinical Trial

E Schreibmann, K Ramesh*, H Shu, H Shim, Department of Radiation Oncology and Winship Cancer Institute of Emory University


(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: Recently, high-resolution whole-brain spectroscopic MRI (sMRI) was proposed to guide radiation dose escalation for glioblastoma patients in a multi-center clinical trial. In this approach, the choline-to-N-acetyl-aspartate (Cho/NAA) ratio is used to define a third, high-risk target volume (GTV3). We report on automation through scripting that incorporates our experience when creating sMRI-based dose-escalated treatment plans.

Methods: Using scripting (ESAPI) we designed an Eclipse plug-in that incorporates our planning strategy, namely choosing the treatment arcs, designing the avoidance structures and selecting the appropriate optimization objectives and weights. GTV3 receiving an escalated dose of 75 Gy is targeted in the same plan with standard volumes of enhancing tumor and non-enhancing and/or edema defined on T1w-MPRAGE and T2-FLAIR, receiving 60 and 51 Gy respectively. To predict the achievable dose to the critical structures, we employ a knowledge-based model. Settings for the optimization and dose calculation options are handled by the script as well and are established by feedback from our dosimetry group. The script generates an automated plan from the approved structures without any user interaction.

Results: All generated plans were useful as an evaluation of the attainable trade-off specific to each patient, that the physician/dosimetrists may tweak in the final review to fit specifics of a particular case. Without any alteration of the automated plans, the mean target dose for (GTV1, GTV2, and GTV3) was (59.9, 68.3, 77.78), comparable to (61.1, 69.6, 77.11) Gy in the clinical plans. Similarly, minimum doses (46.7, 56.2, 70.0), compared to (48.2, 57.1, 69.4). Doses to critical structures were, for example, a mean maximum of 35.1 versus 31.7 for brainstem, 8.1 vs 4.5 for eyes or 37.7 vs 37.5 for hippocampus.

Conclusion: The script synthetizes in one package our department’s trial-and-error experience with these cases to guide the acceptance of this new treatment paradigm into community practice.


Inverse Planning, Operations Research, Spectroscopic Imaging


IM/TH- Informatics: Informatics in Therapy (general)

Contact Email