Dose prediction for cervical cancer VMAT patients with a full-scale 3D-cGAN-based model and the comparison of different input data on the prediction results
Gongsen Zhang, Zejun Jiang, Jian Zhu, Linlin Wang
Abstract
Cervical cancer is the second most common female malignant tumor in the world [ 1 ], for which radiotherapy is currently one of the main treatment methods. Related surveys show that approximately 80% of cervical cancer patients receive radiotherapy at different stages [ 2 , 3 ]. Intensity modulated radiotherapy (IMRT) and volumetric modulated arc radiotherapy (VMAT) have become standard radiotherapy methods. Compared with 3-dimensional conformal radiotherapy (3D-CRT), the dose distribution formed by new technologies above using reverse optimization algorithms is highly consistent with the planned target area and has better uniformity[ 4 , 5 , 6 , 7 ]. However, advanced technology also brings corresponding computational burden, which greatly increases the total planning time. According to statistics, it takes an average of approximately 4 h for radiotherapists to delineate the planning target volume (PTV) and organs at risk (OARs), and may even take longer for some complex diseases. After that, a radiotherapy plan meeting the treatment standards is formulated by radiation physicists, which takes approximately 10 h for each patient [ 8 , 9 ]. The large amount of time required for the treatment plan inevitably leads to delayed treatment, thereby affecting the quality of treatment and prognosis of patients [ 10 ].