Application Research Of Radiotherapy Positioning Based On Augmented Reality Technology

Background:The conventional positioning method based on the combination of laser and body surface markers has many application limitations.For example,the body surface markers are easy to blur or disappear,resulting in interruption of treatment.The positioning verification based on an electronic portal imaging device(EPID)and cone beam computed tomography(CBCT)is not real-time and brings additional radiation dose to patients.The positioning is not intuitive and lacks effective multi-dimensional information management methods.Augmented reality(AR)is an emerging visualization technology,which has been widely studied and explored in the fields of surgical navigation,rehabilitation training,and remote consultation.It has the advantages of integrating virtual and real display,real-time tracking and registration,and intelligent interaction,suggesting its application prospect in guiding radiotherapy positioning.However,the application of AR technology in radiotherapy is still in its infancy,and it is urgent to explore the application value of AR-assisted radiotherapy positioning.Purpose:To develop an AR-assisted radiotherapy positioning system and use AR technology to empower radiotherapy positioning to solve the application limitations in conventional radiotherapy positioning and promote the development of radiotherapy positioning in the direction of non-radiation,real-time correction,visualization,convenience,and intelligence.The accuracy of its positioning for radiotherapy was evaluated by experiments.Methods:The AR-assisted radiotherapy positioning(ARRP)system was developed based on Unity3D and Vuforia SDK and deployed on the Microsoft HoloLens 2(HoloLens 2)device.A cube phantom was used to test the effects of various factors such as distance,angle,illumination,and human interference on the application performance of the ARRP system.Ten tests were carried out under each factor to obtain the optimal application conditions of the system.Based on the optimal application conditions,the positioning of the simulated human phantom was tested 10 times.The CBCT images of the simulated human phantom were obtained after the positioning,and bone registration was performed with its positioning CT images to evaluate the preliminary positioning accuracy of the ARRP system based on HoloLens 2.Two structured light RGB-D(Red-Green-Blue Depth,RGB-D)cameras were introduced to optimize the hardware equipment and improved the ARRP system to further improve the positioning accuracy on the basis of ensuring multi-dimensional information visualization.The positioning of the head&neck,chest,and pelvis of the simulated human phantom was tested 20 times respectively.Under the guidance of the virtual model,the test part of the simulated human phantom was roughly positioned,and then two structured light RGB-D cameras were used to obtain the surface point cloud of the test part of the simulated human phantom in real-time,and the cross-source point cloud registration was performed with the target point cloud generated based on the positioning CT data.The registration result was output to the virtual user interface(UI)in real time to guide the radiotherapy positioning.The cross-source point cloud registration results and the positioning error evaluated by the gold standard CBCT were used to evaluate the positioning accuracy of the improved augmented reality assisted radiotherapy positioning(I-ARRP)system.Pearson correlation analysis was used to evaluate the consistency of the output results of the two positioning guidance methods I-ARRP system and CBCT.Results:In our study,the AR-assisted radiotherapy positioning system was successfully developed and successfully deployed to the HoloLens 2 device.The optimal application conditions of the ARRP system:the registration angle and the tracking angle were consistent and both were 0°,the test distance was 30-50 cm,the light intensity in the treatment room was normal and stable,and no lens occlusion,and no head rotation.The positioning errors of the ARRP system in the Lateral(Lat),Longitudinal(Lng),Vertical(Vrt),and Rtn directions were 3.1 ± 2.0 mm,2.4±2.5 mm,4.6 ± 2.8 mm,and 0.26 ± 0.14°,respectively.The structured light RGB-D cameras were successfully integrated and improved the ARRP system.Using the I-ARRP system,the positioning errors of each test part in the translation directions of Lat,Lng and Vrt were all less than 2mm.Among them,the positioning errors in the Lng translation direction were all less than 1mm,and the positioning errors in the Rtn rotation direction were all 0.2°.There was a linear correlation between the I-ARRP system and CBCT(P<0.001).The Pearson correlation coefficient R in the translation direction of Lat,Lng,Vrt,and the rotation direction of Rtn was greater than 0.8,showing a high correlation,and the positioning guidance was highly consistent.Conclusion:The AR-assisted radiotherapy positioning system based on HoloLens 2 has the advantages of non-radiation positioning verification,intuitive visualization,and convenient interaction,and can provide preliminary accuracy information for precise positioning of radiotherapy.On the premise of retaining the advantages of HoloLens 2 assisted radiotherapy positioning,the structured light RGB-D camera is introduced to optimize the hardware equipment and improve the AR-assisted radiotherapy positioning system,and the positioning accuracy is significantly improved.The AR-assisted radiotherapy positioning system which integrated augmented reality and optical surface imaging technology provides a research basis for exploring a new way of clinical positioning with no radiation,real-time correction,intuitive and convenient.