|
Synthesizing computed tomography images from magnetic resonance images for cervical cancer radiotherapy treatment planning
|
D. Jiang   , Y. Tian , X. Peng , L. Yang , H. Liu |
| Radiotherapy Center, Hubei Key Laboratory of Tumor Biological Behaviors, Hubei Cancer Clinical Study Center, Zhongnan Hospital of Wuhan University, Wuhan, China , yangl_zn@163.com |
|
|
Abstract: (8 Views) |
Background: This study aimed to explore using computed tomography (CT) images synthesized from magnetic resonance (MR) images by the 2.5D semi-supervised generative adversarial nets (SSGAN) framework to enhance the precision of intensity-modulated radiotherapy (IMRT) planning in patients with cervical cancer. Materials and Methods: A comprehensive pelvic MR-CT dataset, encompassing T1-weighted MR and CT volumes from 174 subjects, was utilized. This dataset was divided into training (n = 150) and test (n = 24) sets for model development and assessment. The Hounsfield Unit (HU) discrepancy and dosimetric accuracy of synthetic CT (sCT) images generated by 2.5D SSGAN were evaluated against actual CT images for cervical cancer IMRT planning. Results: The mean gamma analyses values for 2D criteria of 1 mm/1%, 2 mm/2%, and 3 mm/3% in these planes were 92.18% ± 4.64%, 98.13% ± 3.05%, and 99.23% ± 1.52%, respectively. Absolute dose deviations averaged 0.51% ± 0.18% within the regions of interest (ROIs). Conclusion: SSGAN accurately synthesized CT images from MR images, maintaining high dosimetric accuracy essential for cervical cancer IMRT planning. |
|
| Keywords: Magnetic resonance imaging, radiotherapy planning, cervical cancer, Generative adversarial network, synthetic computed tomography. |
|
|
Full-Text [PDF 986 kb]
(2 Downloads)
|
|
Type of Study: Original Research |
Subject:
Radiation Biology
|
|
|
|
|
|
|
| Send email to the article author |
|
|
|
| Add your comments about this article |
|
|
|
