Image Guidance for External Beam Partial Breast Irradiation

Abstract

Patient positioning using laser alignment may provide insufficient accuracy for highly conformal treatments such as partial breast irradiation (PBI). We tested three image guidance methods in patients receiving external beam PBI: 3D video surface mapping, kilovoltage on-board imaging and internal fiducial markers (IFM). The study cohort included patients treated with external beam PBI on an intramural, IRB-approved clinical protocol. Each patient was treated with 9 fractions of 4 Gy over the course of one week. For each treatment, the patients were initially positioned by alignment lasers. The on-board imaging system (OBI, Varian Medical Systems, Palo Alto, CA, USA) was then used to acquire orthogonal images. Projected couch offsets using 2D/2D image matching software were recorded but the couch was not moved. Next, the 3D surface mapping system (AlignRT, Vision RT Ltd, London, UK) was used and the couch was shifted to the final treatment position as specified by automated topographical matching. Verification surface and OBI images were acquired and the recommended couch offsets were recorded but the couch was not further manipulated. Fourteen patients had internal fiducial markers (IFM) placed at the time of surgery. The IFM were identified on orthogonal images of the patient in the final treatment position, and residual setup error against a reference (simulation) image was measured. Twenty-six patients have been studied, and data from 226 treatment fractions were available for analysis. The average absolute shifts for all fractions, analyzed independently, are listed in Table 1. The average three dimensional isocenter shift from the initial setup position using laser alignment was 8.5 mm with AlignRT and 8.4 mm with OBI. There was a striking discrepancy in the recommended treatment isocenters between the two image guidance systems, with an average isocenter difference of 8.4 mm. Using the IFM as a gold standard for accuracy, the residual setup error was less with the AlignRT system compared to the OBI system. The average isocenter shift needed for IFM alignment was 2.9 mm after setup by AlignRT and 7.5 mm after setup by OBI (p = 1.1E-13). Laser alignment alone is sub-optimal in positioning patients for external beam PBI. There is significant discordance in setup recommendations between systems using surface topography and systems using skeletal anatomy. Internal fiducial markers are recommended to provide validation of setup accuracy. As confirmed by internal fiducial markers, 3D surface mapping provides patient positioning accuracy within 2 mm in any one dimension.