Purpose: Online dose verification based on proton-induced positron emitters requires high accuracy in the assignment of elemental composition (e.g. C and O). We developed a machine learning framework for deriving oxygen and carbon concentration based on dual-energy CT (DECT).
Methods: DECT images were synthesized based on digital patient phantoms at the head site using two
methods: 1) theoretical CT numbers with Gaussian noise (method 1) and 2) forward/backward image reconstruction with poly-energetic energy spectrum and Poisson noise added in the projection domain (method 2). Two architectures of convolutional neural networks, UNet and ResNet, were investigated to map from DECT images to C/O weights under different noise levels. Monte-Carlo simulation was employed to identify the impact of fluctuation in oxygen and carbon concentration on activity/dose distribution in proton therapy.
Results: When no additional noise present, all four cases can obtain <2% mean absolute errors (MAE) and <4% root mean square error (RMSE). Both models demonstrate good performance even with the presence of noise. The activity profiles exhibit 3-5% difference in terms of mean relative error (MRE) between the ground truth and the machine learning outcome.
Conclusion: We explored the feasibility of machine learning framework to derive elemental concentration of oxygen and carbon based on DECT images. Compared to conventional methods based on either SECT or DECT, a machine learning framework is advantageous on the following two aspects: geometric prior and noise/artifact immunity. This study lays a foundation for us to apply the proposed approach to clinical DECT images.