ESTRO 38 Abstract book

S609 ESTRO 38

Table 1: measured SNR and observer score Conclusion

Scans with a large WFS setting should not be used for RT delineation of the hippocampus, since shifts of 2mm can be found. Changing from head coil to flex+A&P coils, has more effect on SNR loss than changing to a smaller WFS. Scanning in RT position with flex+A&P coils and low WFS, the 3D TFE scan was considered suitable for delineation. [1] https://www.rtog.org/corelab/contouringatlases/hippoc ampalsparing.aspx PO-1097 Interobserver variability in tumor bed contouring for breast cancer: comparison between RTO and RTT E. La Rocca 1,2 , M. Dispinzieri 1,2 , T. Giandini 3 , V. Lici 2 , S. Frasca 2 , F. Bonfantini 3 , R. Valdagni 1,4 , L. Lozza 2 , E. Pignoli 3 , M.C. De Santis 2 1 University of Milan, Department of Oncology and Hemato-Oncology, Milano, Italy ; 2 Fondazione IRCCS Istituto Nazionale dei Tumori, Radiotherapy Unit 1, Milan, Italy ; 3 Fondazione IRCCS Istituto Nazionale dei Tumori, Medical Physics Unit, Milan, Italy ; 4 Fondazione IRCCS Istituto Nazionale dei Tumori, Director- Radiation Oncology 1 and Prostate Cancer Program, Milan, Italy Purpose or Objective To evaluate the interoperator variability between radiation oncologist (RTO) and radiation therapist (RTT) in tumor bed (TB) contouring for patients (pts) with early breast cancer. Material and Methods We retrospectively analyzed pts undergoing breast conservative surgery who received a boost on the TB. The RTT contoured the TB independently of the RTO after a training course. In the group 1 CTV boost was the surgical bed, defined by adding 1 cm to the surgical clips placed in the lumpectomy cavity during surgery. In the group 2 the boost CTV was defined on the evaluation of preoperative mammography, medical history, intraoperative data and ultimate histological description. Moreover in this group the seroma cavity or a metallic find on the scar were used to identify the CTV boost. The CTVs were three dimension expanded of 0.5 cm creating a planning target volume. We compared the contours in terms of volume, number of slices, and DICE similarity coefficient (DSC). Results Forty pts were evaluated. Twenty pts had surgical clips (group 1), the other twenty had no clips (group 2). For each pts of group 1, no difference in the number of contoured slice was found between the two operators, but a statistically significant difference was found in terms of volumes, being RTT TB on average ~ 45% smaller than RTO TB (9.5±5.5 cm3 vs. 17.4±10.5 cm3). For group 2, random variations between the two operators were found in terms of contour location, number of contoured slices, and volumes, with mean values of 24.7±16.3 cm3 and 26.7±17.1 cm3 for RTT TB and RTO TB, respectively. The TB delineated for this group were significantly bigger (p<0.05) than those delineated by the RTT for group 1. A difference was obtained by comparing the TB volumes delineated by both the operators in each group: 13.4±9.2 cm3 in group1 vs. 25.7±16.5 cm3 in group 2. The mean

Material and Methods A volunteer was scanned on a 3T Philips Ingenia MRI scanner. The 3D T1-TFE sequence had a field of view of 270x253x204mm, 1.1x1.1x1.2mm voxels (FHxAPxRL), TR/TE 6.8/3.1ms, Flip Angle 9 ° , Sense 1.8 (RL). A diagnostic sequence was scanned in head-coil with WFS 1.8 pixel, the optimal RT sequence with flexible and A&P and WFS of 1.0 pixel. Both coil set-up and WFS were varied separately. A separate noise scan without RF was obtained with the same setting and scaling as the 3D T1-TFE. The SNR was obtained by taking the ratio of the mean signal and noise measured in a Region Of Interest (ROI) of 235mm 2 in two locations: a homogeneous part of the ventricle near the hippocampus and the right part of the brain (fig.2). Two trained observers were asked to give the scans a [1- 4] score if the scan was suitable for hippocampus delineation (table 1). Results The SNR measurements are reported in table 1. Changing the WFS from 1.8 to 1.0 pixel results in a 25.9-30.8% decrease in SNR (head coil) and 26.0-26.1% (flex+A&P coils). The use of flex+A&P coils resulted in a reduction in SNR compared to the head coil of 23.1-42.2% depending on location and WFS. The flex+A&P coils give a homogeneous SNR over the image. When scanning in head coil a SNR increase by a factor 1.2-1.3 is seen from center to the right side of the brain. The SNR reduction from the diagnostic to the optimal RT sequence was 46.8% near the hippocampus and 57.1% in the right part of the brain. The observers gave all scans a sufficient score (≥3) for delineation of the hippocampus, irrespectively which type of coils or WFS setting were used.