Room: Exhibit Hall
Purpose: Patient positioning is a critical factor to accurately deliver dose to a treatment target. Patient marks, indexing/immobilization devices, and image guidance assist in localizing the intended radiation target within millimeters of the intended position. The first time the patient is positioned, the treatment table is adjusted until the patient marks are at machine isocenter, and the table position is recorded for future treatments. This study aims to determine the feasibility and accuracy of the prediction of table coordinates prior to treatment. A prior knowledge of table coordinates can prevent wrong site or wrong shift direction errors.
Methods: Baseline table coordinates were obtained by positioning landmark points on the immobilization devices to isocenter. Patient-specific three dimensional coordinates were then predicted by obtaining the patient isocenter position relative to these landmark points in the planning CT. This was not possible in the longitudinal direction for the BodyFix, however, due to a lack of radiographically-apparent landmarks. As a validation, 304 couch coordinates from 100 patients in three dimensions were predicted for on-treatment patients who were immobilized with indexed devices on both Varian and Elekta linear accelerators. These devices include mask with base plate, wingboard, breastboard, and BodyFix. Predicted coordinates were then compared with actual coordinates obtained from treatment validation.
Results: 71% of predicted values were within 1 cm and 86% within 2 cm. The standard deviation of the prediction error was 1.47 cm.The most accurate predictions were from head and neck patients using a base plate (87% within 1 cm) while the lateral direction for breastboard patients had the greatest uncertainty.
Conclusion: Table coordinate prediction is feasible with the consistent use of indexing immobilization devices. A 2 cm action level is appropriate based on the data collected at our institution.