| Objective:Radiotherapy is one of the main treatment methods for head and neck cancer,which can not only preserve the integrity of the anatomical structure of the patient’s organs,improve the quality of life of patients,but also significantly reduce the recurrence of tumors,improve the local tumor control rate,and improve survival.The purpose of this paper is to modify the diagnostic MRI bed by using an MRI bed plate ?and the coil support of high-intensity MRI-compatible composite materials to maintain the same position as CT simulation positioning.The scanned MRI images and diagnostic MRI images are registered with positioning CT images by rigid and deformable registration methods,respectively.To evaluate the geometric and dosimetric differences of gross tumor volume(GTV)and organs at risk(OAR)between different registration methods,and to provide a reference for the application of MRI positioning plate in head and neck tumors.Methods:This study collected 35 patients who received radiotherapy for head and neck tumors at the Oncology Center of Shanxi Bethune Hospital,Tongji Shanxi Hospital from January 2021 to April 2022 and underwent CT simulation and MRI simulation with MRI positioning plate.In addition,these patients also had MRI diagnostic images of the same site.By rigid and deformable registration,we fused the scanned magnetic resonance images of these patients with simulated localization CT images.A total of four sets of images were obtained,named MRI positioning plate_R,MRI positioning plate_D,MRI diagnosis_R,and MRI diagnosis_D.The gross tumor volume(GTV)and organs at risk(OAR)were outlined on these four fused images by the same doctor,and the dose distribution was calculated.The GTV,OAR outlined,and dose distribution on the MRI positioning plate_R images were used as the reference values,and the overlap degree and dose variation of GTV and OAR outlined on the other three sets of images were compared.The geometric evaluation indexes including the Dice similarity coefficient(DSC),Jaccard similarity coefficient(JSC),Hausdorff distance(HD),and Mean distance to agreement(MDA)were used to assess the accuracy of image registration.If the index was a normal distribution,the overall comparison was performed by one-way ANOVA analysis of variance,and the non-parametric Kruskal Wallis rank sum test was used for non-normal distribution.P < 0.05 was considered statistically significant.Results:Compared with the MRI positioning plate_R images,the differences in DSC,JSC,HD,and MDA were statistically significant(P < 0.05)for the GTV,and the rest of the OAR based on the registration mode,except for the brainstem.Compared with the MRI diagnostic images registration,the DSC and JSC of GTV and OAR registration of MRI positioning plate_R and MRI positioning plate_D are higher(DSC > 0.7,JSC > 0.6),HD and MDA were relatively low(HD < 2 cm,MDA < 2 mm),indicating that the volume of GTV and OAR not only overlaps highly in three-dimensional space but also matches well in the degree of edge.With the MRI positioning plate_R prescription dose as a benchmark,the percent volume covered by the remaining three GTV prescription doses changes with less than 5% change,the highest dose changes in the brain stem,left and right parotid glands,spinal cord,and mandible ranged from 0 to 3%.Conclusion:In head and neck tumors,the MRI images obtained by using the MRI positioning plate in the treatment position have high geometric similarity with the positioning CT by rigid registration,which can be used to guide the target area outline of radiotherapy for head and neck tumors.Compared with the registration of MRI diagnostic images and positioning CT,the target dose will be reduced and the radiation dose of organs at risk will be increased.Therefore,for a hospital without an MRI simulator,it has the value of the clinical application. |