ESTRO 2020 Abstract book

S1090 ESTRO 2020

Diagnostic MRI images were routinely registered with the planning CT. Unfortunately, the accuracies of the fusions were compromised due to setup variations, limited FOV and 2D acquisitions with oblique angles. These uncertainties jeopardized the accuracy of target volumes and OAR delineations. Therefore, radiotherapy-specific MRI sequences (RT-MRI) are developed which utilized 3D- isotropic sequences with a scan length similar to the planning CT with minimal distortions and scanned in treatment position with immobilization devices. RT-MRI could replace planning CTs in dose calculations to eliminate registration errors. However, MRI images lacked electron density information which is critical to dose calculations. The solution was to create synthetic CTs from RT-MRI images that contained the appropriate HU information. The purpose of this study was to create synthetic CTs using an atlas-based method and compared it against the planning CT and the resultant VMAT plans derived from the two CTs. Material and Methods The synthetic CT creation used an atlas-based method. First, an atlas was created with 17 rigidly registered RT- MRI and planning CT pairs. Using a leave-one-out approach, the software detected the closest RT-MRI and planning CT pair based on anatomical similarities to the target’s RT-MRI. The two RT-MRIs were registered through deformable registration with manual interventions. The same deformation was applied to the planning CT of that atlas to generate a synthetic CT for the target’s RT-MRI. The two CTs were compared based on mean absolute error (MAE) for all the OARs and dice similarity coefficient (DSC) for the external outline. The VMAT plans were compared by assessing DVH metrics and a 3D gamma analysis.

Conclusion There was definitely potential for this atlas-based method of creating synthetic CTs. With only 17 pairs in the atlas, the results were already promising and comparable to other limited related studies. It was anticipated that results would be improved with a larger atlas. It is definitely possible that atlas-based synthetic CTs could later replace the planning CT for MRI-based planning and then be further incorporated into MRI-only workflows. PO‐1860 Organs‐at‐risk sparing capabilities of Hybrid Partial‐field VMAT during Whole Brain Radiotherapy A.H.L. Yuen 1 , P.M. Wu 2 , A.K.L. Li 1 , P.C.Y. Mak 1 1 St. Teresa's Hospital, Oncology Centre, Hong Kong, Hong Kong SAR China ; 2 The University of Hong Kong, Department of Clinical Oncology, Hong Kong, Hong Kong SAR China Purpose or Objective Although whole brain radiotherapy (WBRT) provides palliation, reduces local recurrence probability and improves overall survival to patients with brain metastases (BMs), it is well documented that the radiation dose to hippocampus is associated with neurocognitive impairment. Recent published clinical study demonstrated that dose to 100% volume (D 100% ) of the hippocampus exceed 9 Gy and maximum dose of the hippocampus exceed 16 Gy in standard WBRT treatment course of 30 Gy/10 fractions were associated with impair memory function. RTOG 0933 phase II trial suggests that minimizing the hippocampus radiation dose to the least possible level may preserve neurocognitive function while not significantly increasing the risk of disease recurrence. In WBRT using conventional double arcs volumetric modulated arc therapy (cVMAT), the large field size creates wide distribution of low dose volume to the hippocampus and other adjacent organs-at-risk (OARs). The present study investigated the potential of a novel technique - hybrid partial-field volumetric modulated arc therapy (HpVMAT) to spare the hippocampus and other OARs during WBRT, and compared with cVMAT. Material and Methods Computed tomography and magnetic resonance images of 10 patients with brain metastases were retrieved in this study. The hippocampus was manually delineated by single radiation oncologist strictly following the RTOG 0933 atlas definition. Plans delivering 30 Gy in 10 fractions were generated for each patient using cVMAT and our novel HpVMAT technique. cVMAT plans comprise 2 coplanar arcs of 359.8° each and collimator rotation of 30° and 330°. Field size was opened up so that the whole brain PTV was completely covered. HpVMAT consists of a pair of static beams and four 180 ° arcs with field size reduction. Collimator angle of the static fields were selected so that both eye balls can be shielded by the X1 collimator jaw. MLC were used to minimize the irradiated OAR (hippocampus) volume. Plan metrics from both techniques

Results In comparison between the CTs, organs with tissue, bone and air interfaces had the greatest errors such as inner ears, mandible and oral cavity. Another area with the noticeable errors were the top of the skull and back of neck due to various head tilt positions. This was evident in the DSC of an average of 0.66. In plan comparisons, there were significant differences in the D max of the head target volume as well as D 95 , D max and D 98 of the neck target volumes. In terms of OARs, there were significant differences in the D mean of the parotids and optic nerves and D max of the lens. For the gamma analysis with constraints of 3%/3mm, the average was 89% in passing rates for the entire body volume.