EP-1800: Setup verification for breast cancer RT: Manual and automatic match of EPID images compared to CBCT

E.L. Lorenzen, K.L. Gottlieb, M. Nielsen
2016 Radiotherapy and Oncology  
Purpose or Objective: Radiotherapy treatments are delivered in our centre using two twin linacs. This provides the possibility of treating patients in either of them. In case of breakdown of one of the linacs, the number of patients interrupting their treatment can be minimised as they can be treated in the linac that continues working. With the aim of optimally doing so, the IGRT workflow is exceptionally changed in case of linac breakdown and image guidance (IG) is only performed when
more » ... formed when considered strictly necessary. Prostate cancer patients treated in our radiotherapy department receive a moderately hypofractionated IMRT treatment with daily IG. The purpose of this work was to assess the dosimetric differences that would result in prostate treatments if IG was not performed in 3, 5 or 7 fractions due to linac breakdown in the most sensitive patients to the lack of IG according to our IGRT protocol. Material and Methods: 20 prostate plans were retrospectively modified and analysed. All of them were moderately hypofractionated treatments with prescription doses to the prostate and seminal vesicle (SV) PTVs of 70 Gy (2.5 Gy/fraction) and 56 Gy (2 Gy/fraction), respectively. They corresponded to patients whose daily positioning shifts after an initial correction of the systematic error showed a standard deviation ≥4mm or an absolute displacement mean value ≥3mm. Seven positioning shifts were randomly selected for each patient out of their recorded treatment data. Beams corresponding to 3, 5 or 7 fractions were accordingly displaced in the TPS, as if no IG had been performed. Results: Dosimetric differences observed for the prostate and SV CTVs were negligible. Mean absolute variations in the mean rectal dose when not performing IG in 3, 5 or 7 fractions were 35.2 ± 27.2 cGy, 50.9 ± 33.8 cGy and 63.2 ± 47.1 cGy, respectively. The results obtained for the bladder were: 19.5 ± 12.9 cGy, 30.0 ± 19.8 cGy and 39.1 ± 31.8 cGy. The table shows the percentage of cases classified by their corresponding absolute variation in the mean dose. Conclusion: This work has been carried out with the data corresponding to the most sensitive patients to the lack of IG. The observed dosimetric effect is greater than the one that would correspond to the mean patient population. In case of exceptionally not performing IG in 3, 5 or 7 fractions due to a breakdown in one of the twin linacs, the prostate and SV CTVs would still be treated correctly with the CTV to PTV margins currently used in our centre. Regarding the organs at risk, the rectum showed the most important dosimetric variations. The dosimetric impact is greater when changing from 3 to 5 fractions without IG than when changing from 5 to 7. Even in this group of patients, the effect of not performing IG in 3 or less fractions would be negligible. Not performing IG in a greater number of fractions could be relevant in cases in which the calculated dose distribution in the rectum is close to its corresponding dose restrictions because further optimisation was not possible. Purpose or Objective: Cone beam computed tomography (CBCT) is generally superior in imaging the patient anatomy due to the 3D representation and the use of kV imaging compared to MV imaging in electronic portal imaging devices (EPIDs). However, EPIDs have the advantage that the treatment fields can be used for the exposure, thereby adding no additional dose to the patient and requiring little additional time. The purpose of this work was to evaluate the use of EPID using both manual an automatic match by comparison to CBCT for setup verification in breast cancer radiotherapy. Material and Methods: Both CBCT and EPID images were acquired in the same patient position for 29 fractions in 10 breast cancer patients. CBCT images were registered automatically to the planning CT using XVI by Elekta based on a grey-value translational match of the thorax wall. EPID images of the medial tangential fields were registered to digitally reconstructed radiographs (DRRs) using either a manual match of the thorax wall by a experienced user using iVIew by Elekta or an automatic match using IGPS by Fratoria. For the EPID registrations the 3D-table corrections were approximated based on the 2D registrations and the beam angle. Results: Bland-Altman plot of the difference in EPID and CBCT registrations is shown in Figure 1 . The mean differences were close to zero for both manual and automatic match of the EPID images. The limits of agreement (1.96 times the standard deviation of the difference) were lower for the manual than the automatic match indicating better agreement with the CBCT. The results of linear regression are shown in Table 1 . The manual match had a higher correlation coefficient (R²) than the automatic match. The match based on EPID generally underestimated the registration obtained by the CBCT as shown by the trend in Figure 1 and by the slope in the regression shown in Table 1 being significantly lower than one.
doi:10.1016/s0167-8140(16)33051-1 fatcat:icgtcpkdgvglhilz5ms64rjama