Ct-Based Mr Synthesis Using Adversarial Cycle-Consistent Networks With Paired Data Learning

Radiotherapy devices using magnetic resonance (MR) imaging are being developed because MR is a safe imaging protocol that provides clear anatomical details. However, the application of MR-based radiotherapy to the aging population is limited because of its high cost and the increased use of metal implants such as cardiac pacemakers and artificial joints. To improve the accuracy of computed tomography (CT)-based radiotherapy planning, we propose a synthetic approach that translates a CT image into an MR image using adversarial cycle-consistent networks with paired data learning. The networks were trained to transform 2D brain CT image slices into 2D brain MR image slices, combining adversarial loss, cycle-consistent loss, voxel-wise loss, and gradient difference loss. The experiments were analyzed using the CT and MR images of 20 subjects, and an ablation study was conducted to show the strength of the proposed objective function. The experimental results show that the proposed method is accurate and robust for predicting MR images from CT images and also outperforms two state-of-the-art methods.