Inter- and intra-fraction geometric errors in daily image-guided radiotherapy of free-breathing breast cancer patients measured with continuous portal imaging

Background. Daily image-guided radiotherapy (IGRT) using two orthogonal setup images may be inaccurate for breast cancer patients treated in free breathing because the setup images may capture the patient in a breathing phase that is not representative of the mean anatomy. The aim of this study was to quantify the setup errors in breast radiotherapy after image-guided setup correction based on two orthogonal setup images acquired in free breathing. Methods and materials. For 16 breast cancer patients with daily image-pair based IGRT, continuous portal imaging (7.5 Hz) were acquired at each treatment fraction during the delivery of the two tangential fields. For each portal image, the chest wall position relative to the planned position was determined in the imager direction orthogonal to the cranio-caudal direction. It yielded the time resolved setup error in this direction throughout the 16 treatment courses. Results. The mean absolute setup error exceeded 5 mm in 0.9% (first field) and 1.8% (last field) of the treatments. The group mean error (M) and the standard deviations of the random (s) and systematic (S) setup errors were M = -0.7 mm, Sigma = 1.1 mm, sigma = 1.5 mm (first field) and M = -0.2 mm, Sigma = 1.4 mm, sigma = 1.7 mm (last field). The negative sign of M indicates that less lung than planned was included in the treatment fields. Intra-field peak-to-peak chest wall motion amplitudes were patient dependent with patient mean values of 2.0 +/- 0.7 mm [ range 1.1-3.2 mm]. The largest observed intra-field motion amplitude was 8 mm. Conclusion. Image-guided setup based on orthogonal planar images acquired in free breathing without synchronization with the respiratory phase was found to result in accurate tangential breast radiotherapy with only few outliers.