Cone-beam CT-based synthetic CT generation for adaptive radiotherapy: methodological paradigms, clinical reliability assessment, and translation
5 affiliations
- 1First Affiliated Hospital of Yangtze University, Yangtze University
- 2Yangtze University Health Science Center, Yangtze University
- 3Beijing Jishuitan Hospital, Capital Medical University
- 4School of Biomedical Engineering, Capital Medical University
- 5Laboratory for Clinical Medicine, Capital Medical University
Abstract
Synthetic computed tomography (sCT) generated from cone-beam computed tomography (CBCT) has important applications in quantitative imaging, dose recalculation, and adaptive radiotherapy. Deep learning has improved image quality and Hounsfield unit accuracy, but its reliability for patient-specific dosimetry and routine clinical use remains uncertain. This narrative review examines deep learning methods for generating sCT from treatment CBCT in photon and proton radiotherapy, with emphasis on data requirements, generalizability, structural fidelity, dosimetric accuracy, and clinical feasibility. The literature search was last updated on 30 July 2026. Current methods mainly include supervised regression, adversarial networks, registrationguided approaches, and diffusion models. Available evidence demonstrates encouraging image and dosimetric performance, although no method consistently performs best across anatomical sites, imaging systems, and treatment modalities. Most studies are based on proof of concept or single-center validation and show considerable heterogeneity in datasets, acquisition protocols, reference images, registration methods, and evaluation criteria. External validation, prospective assessment, failure detection, and quality assurance remain limited. Future studies should emphasize multicenter validation, standardized and modality-specific evaluation criteria, uncertainty assessment, and prospective workflow testing. More rigorous and clinically relevant evidence is needed to support the reliable use of sCT in patient-specific adaptive radiotherapy.
Related
- Opportunities and Challenges in FLASH Radiotherapy: Focusing on Normal Tissue Protection Effects and Clinical Translation Bottleneck
- Editorial: Next-generation radiotherapy for prostate cancer: precision, personalization, and technological advances
- Dynamic Modulated Arc Therapy (DMAT): A Time Aware, Modulation Steered Optimization Framework for Next Generation Radiotherapy Delivery
- Methodological Assessment of Instrumented Indentation for Spatially Resolved Mesoscale Stiffness and Creep in Short Fiber-Reinforced Injection-Molded PBT Components
- Impact of phantom design on fetal radiation exposure assessment in pregnant women undergoing head and neck 6MV photon radiotherapy
BibTeX
@article{zhang2026conebeam,
title = {{Cone-beam CT-based synthetic CT generation for adaptive radiotherapy: methodological paradigms, clinical reliability assessment, and translation}},
author = {Yunxian Zhang and Wanting Zhang and Jinbai Huang and Jingwei Zhao and Huasong Luo and Da He and Zhi Yang},
journal = {Frontiers in Oncology},
year = {2026},
month = oct,
doi = {10.3389/fonc.2026.1981244},
url = {https://www.frontiersin.org/articles/10.3389/fonc.2026.1981244/full},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Cone-beam CT-based synthetic CT generation for adaptive radiotherapy: methodological paradigms, clinical reliability assessment, and translation
AU - Yunxian Zhang
AU - Wanting Zhang
AU - Jinbai Huang
AU - Jingwei Zhao
AU - Huasong Luo
AU - Da He
AU - Zhi Yang
PY - 2026
DA - 2026/10/20
JO - Frontiers in Oncology
DO - 10.3389/fonc.2026.1981244
UR - https://www.frontiersin.org/articles/10.3389/fonc.2026.1981244/full
AB - Synthetic computed tomography (sCT) generated from cone-beam computed tomography (CBCT) has important applications in quantitative imaging, dose recalculation, and adaptive radiotherapy. Deep learning has improved image quality and Hounsfield unit accuracy, but its reliability for patient-specific dosimetry and routine clinical use remains uncertain. This narrative review examines deep learning methods for generating sCT from treatment CBCT in photon and proton radiotherapy, with emphasis on data requirements, generalizability, structural fidelity, dosimetric accuracy, and clinical feasibility. The literature search was last updated on 30 July 2026. Current methods mainly include supervised regression, adversarial networks, registrationguided approaches, and diffusion models. Available evidence demonstrates encouraging image and dosimetric performance, although no method consistently performs best across anatomical sites, imaging systems, and treatment modalities. Most studies are based on proof of concept or single-center validation and show considerable heterogeneity in datasets, acquisition protocols, reference images, registration methods, and evaluation criteria. External validation, prospective assessment, failure detection, and quality assurance remain limited. Future studies should emphasize multicenter validation, standardized and modality-specific evaluation criteria, uncertainty assessment, and prospective workflow testing. More rigorous and clinically relevant evidence is needed to support the reliable use of sCT in patient-specific adaptive radiotherapy.
SN - 2234-943X
ER -