Igrt With Ct-On-Rails For Prone Breast Irradiation

Accurate and reproducible patient positioning in partial breast irradiation has been challenging particularly for prone breast setup. In this work, we explore the use of high image quality kV CT to improve the accuracy and reproducibility in prone breast patient set-up. An in-house study that enrolls patients with prone breast irradiation to be treated on linac and CT-on-Rails combination (CTVision, Siemens) is ongoing. Patients who could fit through the standard bore of the CT with the field of view sufficient to image the whole treated breast and had surgical clips placed in the lumpectomy cavity at time of surgery are selected. The CT imaging was acquired for 5 consecutive fractions. The patients were set up prone with shifts made from a PA set up point to the isocenter. The treatment table was then rotated 180o and the patient was scanned by the CT. The isocenter of the daily scan was found using BBs placed during set up. The scan was automatically registered based on bony anatomy with the treatment planning CT. Registration was adjusted manually to achieve visual agreement between the daily and planning scan using the surgical cavity or at least 3 surgical clips. The patient was repositioned based on the determined shifts. Orthogonal portal images were taken prior and after patient repositioning. All the films prior to the shifts were approved based on the chest wall position. Each lateral film was overlaid with DRR contours and aligned based on soft tissue agreement. The difference between the DRR isocenter and the treatment isocenter (delineated by the tungsten rod tray inserted during imaging) were noted. The shifts determined from portal images were compared to the respective shifts based on CT registration. Four patients have so far gone through the study. The IGRT procedure is well tolerated. Reflecting the difficult nature of the patient set up, rotation of the torso and rotation/distortion of breast tissue are easily seen in the CT images, and can be corrected if necessary. The distortion of the breast tissue affects position of the clips and surgical cavity. Manual registration is necessary to achieve the best inclusion of the cavity in the target volume. The magnitude of the daily shifts varied for each of the patients, with the smallest average shift of 0 ± 3mm and the largest of -18 ± 3mm. The shifts determined from CT registration and from portal images were strongly correlated (r = 0.8 for vertical and r = 1.0 for longitudinal directions).The clips were not identifiable on the portal films, and could not serve as surrogates for cavity identification. The large patient variability and the interfractional variability are indicative that a 3D image guidance with high soft tissue contrast, such as the kV fan beam CT, is necessary for accurate and reproducible treatment delivery in prone breast treatment.