Back to all papers

Novel extended field-of-view reconstruction algorithms in computed tomography simulation for imaging and volumetric modulated arc therapy dosimetry.

May 18, 2026pubmed logopapers

Authors

Piao Z,Wang G,Wang Y,Wei Z,Qi M,Ding S,Song H,Feng Y,Jia L,Sun H,Huang S,Huang X

Affiliations (2)

  • State Key Laboratory of Oncology in South China, Guangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and Therapy, Guangdong Provincial Clinical Research Center for Cancer, Sun Yat-sen University Cancer Center, Guangzhou, China.
  • Research Cooperation Department, United Imaging Research Institute of Innovative Medical Equipment, Shenzhen, China.

Abstract

Accurate computed tomography (CT) reconstruction is essential for effective radiotherapy, particularly when the patient anatomy extends beyond the standard scanning field of view (sFOV). Extended-FOV algorithms address this challenge, but their performance in clinical CT simulations remains unclear. This study systematically evaluated two algorithms, the extended-FOV reconstruction algorithm (eFOV) and the deep learning-enhanced eFOV (AI-eFOV), on the uCT610 Sim system (United Imaging Healthcare), with CT image quality and volumetric modulated arc therapy (VMAT) dosimetric accuracy being the outcomes. A whole-body phantom simulated off-center displacements (5-20 cm) for extended FOV reconstruction. Isocentric CT images served as the ground truth for the assessment of geometric distortion, Hounsfield unit (HU) stability, and autocontouring accuracy. VMAT plans for the lung, chest wall, and liver were analyzed for dosimetric impact, and four patient cases provided clinical validation. With increasing displacement, eFOV showed large HU deviations [root mean squared error (RMSE) 96 HU; maximum error 275 HU for bone and 293 HU for soft tissue], whereas AI-eFOV reduced errors (RMSE 81 HU; maximum 65 HU for bone and 25 HU for soft tissue; P<0.01) and achieved higher autocontouring accuracy (Dice ≥0.973 for AI-eFOV and ≥0.926 for eFOV). In VMAT plans, both algorithms achieved >96.66% gamma passing rates (3%/3 mm) within a 10-cm offset. At 20 cm, AI-eFOV maintained a rate above 93.37%, while eFOV dropped to 82.05% in hypofractionated plans using the flattening-filter-free high dose rate delivery mode. Clinical data further confirmed AI-eFOV's advantage in artifact suppression and tissue boundary delineation. Image distortion is determined by anatomy and lateral displacement beyond the sFOV. AI-eFOV provided superior image quality and dosimetric accuracy, especially beyond 70 cm and should be preferred for cases with large offsets.

Topics

Journal Article

Ready to Sharpen Your Edge?

Subscribe to join 11k+ peers who rely on RadAI Slice. Get the essential weekly briefing that empowers you to navigate the future of radiology.

We respect your privacy. Unsubscribe at any time.