Exploratory development and clinical research of a mixed reality guided radiotherapy positioning system

Purpose

This study aims to develop an application software deployed on HoloLens2 for guiding patient radiotherapy setup and assessing its feasibility and accuracy in radiotherapy.

Methods

The developed system features an application software deployed on the HoloLens2 device, enabling automatic tracking and registration of a virtual cube to the location of a real calibration cube. This functionality is achieved through the utilization of the Vuforia SDK, which determines the location point linking the virtual and real spaces. The system then generates a reference hologram based on simulated positioning computed tomography (CT) data sets and anchors this reference hologram at the determined location point. In the process of treatment positioning, the real human body in the scene of HoloLens2 is matched with the surface of the reference hologram to complete the positioning. This paper explores the factors that affect the accuracy of the system, namely the viewing distance and Angle, to determine the optimal conditions for using the system. Then, based on the optimal conditions, the error between the cone beam computed tomography (CBCT) data of patients (head, neck and chest and abdomen) obtained by the system and conventional positioning were compared by T-test method to evaluate the performance of the system.

Results

The observation distance had a minor impact on the system's accuracy, as there was no significant difference between the average six-dimensional error and the gold standard (all P > 0.05). A significant impact on the positioning error of the system was observed when the observation angle and isocentric deviation were both set at 45° (the vertical error was notably different from the gold standard, P < 0.05). The observation angle (0°–30°) exhibited a slight influence on the accuracy of the system (all P > 0.05). Patients with head and neck cancer exhibited a significant difference in positioning error between this system and the conventional method in the vertical direction (1.0 ± 0.9 mm vs. 1.8 ± 0.8 mm, P < 0.05), while no statistical significance was noted in other directions. Patients with chest and abdominal cancer showed significant discrepancies in the transformation between the two systems in lateral direction (1.9 ± 1.0 mm vs. 2.5 ± 0.7 mm, P < 0.05), vertical direction (1.4 ± 1.2 mm vs. 2.2 ± 1.1 mm, P < 0.05), and longitudinal direction (2.0 ± 1.2 mm vs. 3.1 ± 1.4 mm, P < 0.05), with no statistical significance found in rotation.

Conclusion

In this study, a radiotherapy treatment positioning system utilizing mixed reality guidance is introduced. This system demonstrates the capability to notably decrease positioning errors in the transformation of patients with chest and abdominal cancer, as well as in the vertical direction for patients with head and neck cancer. Furthermore, it shows enhancements in the overall accuracy of radiotherapy procedures.