ESTRO 2021 Abstract Book

S1624

ESTRO 2021

basis of the overall plan quality and also partial score of each OAR. Wilcoxon’ rank test was used to spot statistically significant differences and Pearson’s correlation coefficient was also computed. To assess the capability of each algorithm to act as a classifier for the quality of the other center’s plans a ROC analysis has been performed. Results

Table 1 depicts the PQM algorithms and their component. Center B devotes more attention to the targets with respect to the Center A, to the expense of the relative weights of bladder and femoral heads. This difference directly reflects on the overall PQM% score, as can be seen in Table 1. The overall score resulted non-correlated and significantly different. Such behavior is predominantly due to the difference in the PTV scores while the scores for CTV, bladder and rectum are all well correlated. The relevance of each classifier was measured with AUC, which resulted 0.833 [0.503;0.998] when the algorithm of Center A is used to classify plans from Center B and 0.798 [0.512;0.967] when switching the roles. An example of the ROC curve is given in Figure 1 for different levels of acceptability.

Conclusion Two PQM algorithms independently created from two different centers were compared. Their scoring was applied to a sample of patients coming from both centers. While OAR scoring resulted well correlated, marked difference and an absolute lack of correlation was found in PTV scoring. The target coverage and overdosage are strictly linked to the clinical practice of each institution while the levels of OAR sparing seems to show a more homogenous consent.

PO-1904 Commissioning HyperArc for Single Targets including Benign Tumours D. Egleston 1 , R. Brass 1 1 The Clatterbridge Cancer Centre, Physics, Liverpool, United Kingdom

Purpose or Objective To show HyperArc™ plans have equivalent or improved plan quality compared to RapidArc™ VMAT, for SRS treatment of brain metastases, meningiomas, and acoustic neuromas, with the overall goal of commissioning HyperArc™ treatments for

single cranial targets. Materials and Methods

8 patients who had previously received non-coplanar RapidArc™ VMAT for single targets on a Varian Novalis Tx linac were replanned using HyperArc™ for treatment on a Varian Edge linac. The patients represented a range of sites within the skull, constituting 4 brain metastases, 2 acoustic neuromas, and 2 meningiomas. 4 plans had isocentres within the central patient protection zone, and were planned with both full arc geometry using the standard 5 half-arcs and reduced arc geometry using 4 half-arcs excluding the central arc as in Figure 1. The other 4 had lateral isocentres, and were planned with reduced arc geometry of 4 half-arcs excluding the contra-lateral arc. 1 brain metastasis and 1 meningioma patient additionally had some overlap of the PTV with brainstem. Plan quality was compared to the previous RapidArc™ VMAT plan. The effects of normalising to deliver 80% dose to 99% of the PTV were additionally investigated. Each HyperArc™ plan was verified by analysis with both PDIP and Octavius, and analysed by DoseCHECK™ independent dose calculation.