Computer-controlled non-invasive patient positioning in fractionated radiotherapy - a videogrammetric system for automatic patient set-up, fast detection of patient motion and online correction of target point misalignment during therapy

Purpose An integrated videogrammetry based system for high precision fractionated radiotherapy in head and neck region has been developed. Patient setup is performed automatically and patient motion during treatment is detected and compensated. Methods and Materials The system consists of two video cameras controlled by a personal computer, a landmark system with a device for dento-maxillary fixation (DMF) and a computer-controlled motorized tabletop. Individually shaped, the dento-maxillary fixation fits the patient's teeth of the upper jaw. The optical measurement system is calibrated relative to the isocenter of the therapy unit. Position and spatial orientation of the landmarks are monitored by the cameras and processed for the real-time calculation of a target point's actual position relative to its initial position. Calculated translational misalignments are compensated online by countermovements of a 3D computer-controlled tabletop. Rotational deviations cannot be corrected automatically, but non-isocentric rotations are separeted into isocentric rotations and translations, which are online corrected. Thus the planned target point remains in the isocenter. Results System inaccuracies due to algorithm, calibration, illumination dependence, scale factors for image formation, have been determined in phantom studies. Algorithm and calibration errors are smaller than 0.2 %. Illumination dependence is uncritical for the infrared emitting diodes (IRED) used. Spectrally adapted bandpass filters eliminate interferences, caused by other light sources. The DMF repositioning accuracy has been derived from clinical studies on 11 patients to be 0.280±0.140 mm (mean±std. dev.). Spatial resolution of the optical measurement system is 5 μm for translational and 20 μrad for rotational deviations with a sampling frequency of 15–20 Hz. The maximum speed of the tabletop is 25 mm per second, the spatial resolution is about 10 μm with a default residual motion stop of 50 μm for translational deviations with respect to the initial targetpoint position. In summary, the overall positioning accuracy of the complete system was determined to be 0.6 mm. Conclusion For stereotacically guided radiotherapy in head and neck region the developed patient positioning sensor unit (PPSU) combines the high positioning accuracy, known from invasive stereotactic fixation in single-fraction radiotherapy, together with well known biological advantages for a fractionated regimen. The high positioning accuracy allows to decrease the commonly added safety margins. This leads to an enormous reduction of irradiated tumor-surrounding healthy tissue. The PPSU realizes automatic patient setup, patient motion detection and online movement compensation during therapy.