LAUSR

PURPOSE The purpose of this study is to investigate the effect of different MR sequences on the accuracy of deep learning based synthetic CT (sCT) generation in the complex head and neck region. METHODS Four sequences of MR images (T1, T2, T1C and T1DixonC-water) were collected for 45 patients with nasopharyngeal carcinoma. Seven conditional generative adversarial network (cGAN) models were trained with different sequences (single channel) and different combinations (multi-channel) as inputs. To further verify the cGAN performance, we also used a U-net network as a comparison. Mean absolute error, structural similarity index, peak signal-to-noise ratio, dice similarity coefficient, and dose distribution were evaluated between the actual CTs and sCTs generated from different models. RESULTS The results show that the cGAN model with multi-channel (i.e., T1+T2+T1C+T1DixonC-water) as input to predict sCT achieves higher accuracy than any single MR sequence model. The T1-weighted MR model achieves better results than T2, T1C and T1DixonC-water models. The comparison between cGAN and U-net shows that the sCTs predicted by cGAN retains additional image details, are less blurred and more similar to the actual CT. CONCLUSION cGAN with multiple MR sequences as model input shows the best accuracy. The T1-weighted MR images provide sufficient image information and are suitable for sCT prediction in clinical scenarios with limited acquisition sequences or limited acquisition time.