Feasibility of improving daily MVCT image quality using RayStation-generated synthetic CT for enhanced adaptive radiotherapy applications: a phantom study

Abstract

This study investigated whether synthetic CT (sCT) generation within the RayStation treatment planning system could improve the image quality of daily low-dose megavoltage CT (MVCT) from a TomoTherapy system (Accuray Inc., Sunnyvale, CA, USA). sCT images were generated from paired kilovoltage computed tomography (kVCT) and MVCT datasets acquired from a tissue-equivalent phantom using deformable registration based on the ANAtomically CONstrained Deformation Algorithm (ANACONDA) and Hounsfield unit (HU) correction. Although originally developed for cone beam CT (CBCT) enhancement, this technique was applied to MVCT in this study. Quantitative image analysis showed that the sCT images preserved the geometric alignment of MVCT while achieving HU accuracy, contrast, and spatial resolution comparable to those of kVCT. Based on structural similarity and pixel-level metrics, sCT demonstrated reduced distortion and improved geometric fidelity compared to MVCT, achieving, for example, a Structural Similarity Index Measure of 0.902 with kVCT and a contrast-to-noise ratio approximately 3.4 times higher than MVCT. These results support the feasibility of applying RayStation's sCT generation functionality to MVCT, extending its utility beyond CBCT. This approach may contribute to improved image guidance and adaptive planning accuracy in TomoTherapy workflows, providing a foundation for future clinical investigations.