Fig. 1

Framework of the proposed AGGAN, which contains two attention-guided generators G_CBCT–CT and G_CT–CBCT. We show one cycle in this figure, i.e., CBCT → sCT → rCBCT ≈ CBCT. Each generator such as G_CBCT–CT consists of a parameter sharing encoder E_CBCT–CT, a content mask generator \(G_{CBCT - CT}^{C}\) and an attention mask generator \(G_{CBCT - CT}^{A}\). The proposed model is constrained by the cycle-consistency loss. The symbols ⊕ ,  ⊗ and Ⓢ denote element-wise addition, element-wise multiplication and channel-wise Softmax respectively