ESTRO 36 Abstract Book

S558 ESTRO 36 _______________________________________________________________________________________________

Material and Methods The recently introduced DR protocol was enrolled on 7 patients, counting for 206 CBCT-planning CT (pCT) registrations. All 3D rotations and translations, together with categorical data were recorded. (e.g. Did scan lead to treatment? , Did DR algorithms perform well?, Were predefined limits exceeded? Final approval (RTT /MD)? Categories for deviation from pCT: pelvic tilt, bladder filling, tumor shrinkage,…). Based on this DR experience, a TLP dedicated for LACC pre-treatment registration was set up in collaboration with an RTT, Physician (MD) and Physicist. The MD and RTT applied the TLP to 30 scans (1 patient) as an initial validation. A 4-scale score reflecting the degree of confidence about the RTTs decision was taken up. Results The necessity of the DR protocol, in comparison with a CVOI bony anatomy AR alone, was emphasized by the absolute mean (and maximum) differences of 0.2 (1.5) cm (L/R), 0.3 (2.9) cm (S/I) and 0.3 (1.5) cm (A/P) between the two methods. In 7.3% of the cases the DR was misled due to high contrast regions in the intestine (alternated gastrografine and air bulbs), requiring manual adjustments. For 19.6 % of the CBCT scans that finally led to treatment and where DR performed well, anatomical deviations had been reported and not entirely corrected for (table 1). Some are: pelvic tilt > 3°, tumor shrinkage, bladder or rectum over- or underfilling, or exceeding of translational DR limits. In 25.2 % of all CBCT scans, MD support for final decision was needed. The color code of the TLP is green= go ahead, orange = go ahead AND e-mail to attending MD AND copy note into Record &Verify system; red = do not proceed and call support of MD. In this way reporting was standardized, tumor shrinkage is followed more strictly in consideration of re-planning, and DR failure led to clear recommendations for manual registration. The TLP’s results were validated to be ‘correct’. The confidence level of the RTTs on their final decision, after following the TLP was high (figure 1).

3x18Gy(Gp2) database. Patient characteristics were compared and intra-fraction tumor position variability in Left/Right(LR), Cranial/Caudal(CC) and Anterior/Superior AP) was calculated in terms of Group Mean(GM), systematic(∑) and random(σ) variations for both groups from either the 1 st intra-arc CBCT or a CBCT taken between the arcs if no intra-arc CBCT was available . Results No significant difference in patient characteristics between the groups was observed; gender: female 44% vs 48%, male 56% vs 52%, median age 78.5 vs 72.0years and mean GTV 32.2 vs 12.3cc in Gp1 and Gp2 respectively. Intra-fraction tumor position variability in both groups was small in all directions. The GM and random error were not significantly different between the 2 groups whereas the systematic error was significantly smaller for the 8x7.5Gy group in the AP direction (table 1). Twelve (75%) patients completed the new protocol i.e. no validation scan throughout the whole of their treatment. Four (25%) patients started with a no validation protocol but reverted to the standard SBRT IGRTprotocol after a varied number of fractions (fraction 2-5) when the threshold of ≥0.3cm was exceeded in the 1 st intra-arc scan. For some patients the intra-arc CBCT was not acquired due to technical issues or intra-arc CBCT image quality and a CBCT between the arcs was used for verification. This occurred in 35% of the fractions. Besides omitting the verification scan some patients (9% fractions) required an intervention between arcs.

Conclusion The setup correction validation scan could be safely omitted for most patients with central tumors treated with a 8x7.5Gy dose regimen. Nevertheless, careful monitoring is recommended for any SBRT dose regimen to capture patients with larger intra-fraction position variability. PO-1012 Traffic Light Protocol as a guide for optimal registration of LACC complex tumor pathology E. Bogaert 1 , L. Van den Berghe 1 , A.L. Michiels 1 , C. De Wagter 1 , Y. Lievens 1 , K. Vandecasteele 1 1 Ghent University Hospital, Radiation Oncology, Gent, Belgium Purpose or Objective Pre-treatment CBCT imaging for locally advanced cervix carcinoma (LACC) is challenging. A Dual Registration (DR) protocol (Mask-based VOI (MVOI) Autoregistration (AR) for the primary PTV + Clipbox-Based (CVOI) AR for lymph node (LN) coverage) had been set up with limits of 3 °, 6 mm for CVOI and 3 mm for MVOI AR. However DR optimizes target and LN coverage, it does not provide a decision framework in case anomalies occur. The purpose is to setup a Traffic Light Protocol (TLP) for RTTs, that includes DR and where recommendations for Treat, Reposition or Call Support are clearly outlined and specific actions to take are visualized in green/orange/red.

Conclusion For CBCT pre-treatment imaging of LACC, Clipbox based bony AR alone is less accurate than DR. Even so DR alone is not sufficient. A TLP with DR included was setup to improve the level of correct and fast decision making. In this way, RTTs felt confident about their decision.