Low-Dose Cone Beam CT Reconstruction by Deep Neural Network for Image-Guided Radiation Therapy

Radiation therapy is regarded as the mainstay treatment for cancer in clinic. Kilovoltage cone-beam CT (CBCT) images have been acquired for most treatment sites as the clinical routine for image-guided radiation therapy (IGRT), but repeated CBCT scanning brings extra irradiation dose to the patients. To decrease the extra dose while maintaining the quality of CBCT, deep learning methods are widely adopted. In this study, planning CT (pCT) was introduced to construct a strictly structure-preserved CBCT as prior information. Based on generative adversarial network (GAN), a deep-learning ultra-sparse strictly structure-preserved (DUSS) CBCT reconstruction model was established. Extreme low sampling rates were used for rapid reconstruction of CBCT with high quality and its clinical performance in head-and-neck cancer patients was evaluated.