Assessment of Bone Electron Density Effects on Dose Calculation and Optimization Accuracy for MRI-Only Treatment Planning for Cervical Carcinoma in 1.5 T MR-Linac

Abstract

For cervical patients planning, conventionally only part of the bony regions like femoral head will be contoured, which can lead to the misinterpretation for the undelineated bones on the bulk electron density (ED) assigned MRI-only planning. This study aims to assess the impact of these undelinated bone regions on dose calculation and optimization results of MRI-only based cervical IMRT plans. Five retrospective cervix patients who were treated on the 1.5T MR-Linac with their original CT and MRI datasets were used for the study. Both of the IMRT plans on CT and MRI images were created using TPS Monaco 5.40. After registration of the MRI with the original CT, the synthetic CT (s-CT) for MRI was generated using bulk ED assignment according to the mean ED values of the delineated structures from the original planning CT. To investigate the dosimetric impact of bone electron density on MRI-only planning, another set of s-CT was generated with fully delineation of all the bone regions. The adaptive plans for the two s-CT sets, which denote as “BoneRC-MRI” and “NoBoneRC-MRI” respectively, were created by copying from the original CT and recalculating final dose without optimization. To account for the influence from planning optimization process, the above s-CT based adaptive plans were copied and reoptimized using the method of optimizing shapes, which denote as “BoneRO-MRI” and “NoBoneRO-MRI”. All the adaptive plans were recalculated on the original CT for comparison. The dose-volume-parameters (DVPs), Homogeneity Index, Conformity Index and Gamma analysis were implemented for assessment. In comparison of the recalculation plans on two s-CT sets, NoBoneRC-MRI plans significantly increase PTV dose (Dmean: BoneRC-MRI vs NoBoneRC-MRI, p＜0.001). Comparing the recalculation plans on original CT and s-CT, the result showed that the BoneRC-MRI plans lead to smaller dose effect than NoBoneRC-MRI by comparing RC-CT, with the difference being 0.7% and 1.2% respectively. And gamma pass rate (GPR) of BoneRC-MRI was higher than NoBoneRC-MRI (BoneRC-MRI vs NoBoneRC-MRI, p = 0.049). In addition, comparing the reoptimized plans on two s-CT sets, the DVPs were no significantly difference (PTV Dmean: BoneRO-MRI vs NoBoneRO-MRI, p = 0.572). However, in comparison of reoptimized plans on original CT and s-CT, the result showed that the BoneRO-MRI plans also lead to smaller dose effect than NoBoneRO-MRI by comparing BoneRO-CT and NoBoneRO-CT, with the difference being 0.7% and 1.2% respectively. And the GPR of BoneRO-MRI was higher than NoBoneRO-MRI (BoneRO-MRI vs NoBoneRO-MRI, p = 0.008). The commonly used bulk ED assignment with only part of bone delineation may not be sufficient for cervical cancer MRI-only planning. Fully delineation of the bones and considering the effect of bone electron density have the potential to reduce the error of s-CT generation.