Surface imaging for real-time patient positioning in external radiation therapy

Purpose In the last few years, there has been a growing interest in surface imaging for patient positioning in external radiation therapy. The aim of this study is to evaluate the accuracy of daily patient positioning using the Azure Kinect surface imaging. Methods A total of 50 fractions in 10 patients including lung, pelvic, and head and neck tumors were analyzed in real time. A rigid registration algorithm, based on the iterative closest point (ICP) approach, is employed to estimate the patient position in 6 degrees of freedom (DOF). This position is compared to the reference values obtained by the radiograph imaging. The mean setup error and its standard deviation were calculated for all measured fractions. Results The positioning error showed 1.1 +/- 1.1 mm in lateral, 1.8 +/- 2.1 mm in longitudinal, and 0.8 +/- 1.1 mm in vertical, and 0.3 degrees +/- 0.4 degrees in yaw, 0.2 degrees +/- 0.2 degrees in pitch, and 0.2 degrees +/- 0.2 degrees in roll directions. The larger setup error occurred in pelvic regions. Conclusion We have evaluated in a radiotherapy set-up considering different patient anatomical locations, a depth measurement based surface imaging solution for patient positioning considering the 6 DOF couch motion. We showed that the proposed solution allows an accurate patient positioning without the need for patient markings or the use of additional radiation dose.