Iranian Journal of Medical Physics

Iranian Journal of Medical Physics

Development and Validation of a TPS-Derived Feature Based Predictive Model for Adaptive Radiotherapy Support: APE-Net

Document Type : Original Paper

Authors
daniel Warurii Ndegwa 1
Martin M Mutie 2
Paul K Kamweru 3
James Ngure 4
1 P.O. BOX 109-60400 Chuka, KENYA
2 P.O. Box 109 – 60400 Chuka P.O. Box 109 – 60400 Chuka
3 Department of Physical Sciences, Chuka University
4 Department of Radiotherapy, Nakuru County Referral and Teaching Hospital
10.22038/ijmp.2026.95287.2688
Abstract
Introduction: Radiotherapy for advanced head and neck cancer (HNC) is complicated by anatomical changes, tumor shrinkage, and organ displacement, which may cause dosimetric inaccuracies and increased toxicity. Adaptive radiotherapy (ART) remains challenging in resource-limited settings due to time and personnel constraints. This study investigated the feasibility of the Anatomy-Preserving Estimation Network (APE-Net), a treatment planning system (TPS)-informed predictive framework for supporting offline adaptive radiotherapy (oART).
Material and Methods: Data from 21 retrospectively analyzed HNC patients were used, including baseline (Plan 1), unadapted (Plan 2), and adapted (Plan 3) treatment plans. Patients underwent cone-beam computed tomography (CBCT) monitoring, with rescan CT (rCT) acquired when anatomical deviations of >1 cm occurred. Plan 2 was recalculated using rCT images and original treatment parameters, while Plan 3 was re-optimized using the Eclipse TPS (Version 16, Varian Medical Systems, Palo Alto, CA, USA) at Moi Teaching and Referral Hospital, Eldoret, Kenya. TPS-derived dosimetric and volumetric features were extracted. 108 synthetic TPS-derived feature samples for augmentation were generated. APE-Net was implemented as a fully connected multilayer perceptron (MLP) trained on TPS-derived feature vectors to estimate adaptive treatment parameters.
Results: APE-Net trained on both real and synthetic TPS-derived features demonstrated moderate predictive performance, achieving a coefficient of determination (R²) of 0.681, mean absolute error (MAE) of 9.092% and root mean squared error (RMSE) of 16.776%. Training and validation loss curves showed stable convergence with limited divergence.
Conclusion: APE-Net demonstrates feasibility as a TPS-informed machine-learning framework for supporting predictive oART workflows in resource-limited settings.
Keywords
Head and Neck Cancer (HNC) Limited
Resource Setting Synthetic TPS
Derived Feature Predictive Modelling Anatomical Changes
Subjects
1.        Elizaveta Lavrova, Garrett MD, Wang Y, Chin C, Elliston CD, Savacool M, et al. Adaptive Radiation Therapy: A Review of CT-based Techniques. Radiology. 2023 Jul 1;5(4).
2.        Morgan HE, Sher DJ. Adaptive radiotherapy for head and neck cancer. Cancers of the Head & Neck. 2020 Jan 9;5(1).
3.        Maja Guberina, Nika Guberina, Hoffmann C, A. Gogishvili, Freisleben F, Herz A, et al. Prospects for online adaptive radiation therapy (ART) for head and neck cancer. Radiation oncology. 2024 Jan 8;19(1).
4.        Zafar F, Vilsan J, Mani S, Al AR, Cano-Reyes SE, Abraham G, et al. AI in radiation oncology: a comprehensive review of current applications and future directions. Cureus [Internet]. 2025 Sep 22 [cited 2025 Sep 23];17(9):e398228.
5.        Blumenfeld PA, Feldman J, Arbit E, Weizman N, Berger A, Hillman Y, et al. Daily Artificial Intelligence-Assisted Adaptive Radiotherapy on Cone-Beam CT for Cancer of the Head and Neck. International Journal of Radiation Oncology*Biology*Physics. 2022 Oct 22;114(3):e590–1.
6.        Liu H, Schaal DW, Curry H, Clark R, Magliari A, Kupelian PA, et al. Review of cone beam computed tomography based online adaptive radiotherapy: current trend and future direction. Radiation Oncology. 2023 Sep 2;18(1).
7.        Glide-Hurst CK, Lee P, Yock AD, Olsen JR, Cao M, Siddiqui F, et al. Adaptive Radiation Therapy (ART) Strategies and Technical Considerations: A State of the ART Review From NRG Oncology. International Journal of Radiation Oncology*Biology*Physics. 2021 Mar;109(4):1054–75.
8.        Chen L, Sun H, Wang Z, Zhang T, Zhang H, Wang W, et al. Deep learning architecture with shunted transformer and 3D deformable convolution for voxel-level dose prediction of head and neck tumors. Physical and Engineering Sciences in Medicine. 2024 Aug 5;47(4):1501–12.
9.        Subramanian S. Training and Inference Time Dynamics of Artificial Neural Networks. Artificial Intelligence. 2025 Nov 14.
10.     Temmar H, Willsey MS, Costello JT, Mender MJ, Cubillos LH, DeMatteo JC, et al. Investigating the benefits of artificial neural networks over linear approaches to BMI decoding. J Neural Eng [Internet]. 2025 Jun 17 [cited 2025 Sep 20];22(3):036012.
11.     Wang L, Jiang J, Ghita O, Zhu Y, Sucala VI. Knowledge-enabled stress–strain prediction in additive manufacturing via transferable deep temporal modelling. Advanced Engineering Informatics. 2026 Jan;69:104004.
12.     Montesinos López OA, Montesinos López A, Crossa J. Overfitting, model tuning, and evaluation of prediction performance. In: Multivariate statistical machine learning methods for genomic prediction [Internet]. Springer; 2022. p. 109–139.
13.     Shaban OS, Omoush A. AI-Driven Financial Transparency and Corporate Governance: Enhancing Accounting Practices with Evidence from Jordan. Sustainability. 2025 Apr 23;17(9):3818.
14.     Chai T, Draxler RR. Root mean square error (RMSE) or mean absolute error (MAE)? – Arguments against avoiding RMSE in the literature. Geoscientific Model Development. 2014 Jun 30;7(3):1247–50.
15.     Vandewinckele L, Claessens M, Dinkla A, Brouwer C, Crijns W, Verellen D, et al. Overview of artificial intelligence-based applications in radiotherapy: Recommendations for implementation and quality assurance. Radiotherapy and Oncology [Internet]. 2020 Dec 1;153:55–66.
16.     Song B, Jiang G, Liu Y, Cheng D. A clinical prediction model for central venous catheter-associated bloodstream infections: derivation and validation from the MIMIC-IV cohort. European Journal of Medical Research. 2026 Jan 6;31(1).
17.     Guzman-Pando A, Enriquez-Guillen BO, Ramirez-Alonso G, Camarillo-Cisneros J, Aguilar-Torres CR, Hinojos-Gallardo LC. AI-FRS: An Ensemble-Based AI Decision-Support System for Fetal Risk Prediction in a Mexican Clinical Setting. AI. 2026 Apr 1;7(4):129.
18.     Zhang Z, Zhao Y, Canes A, Steinberg D, Lyashevska O. Predictive analytics with gradient boosting in clinical medicine. Annals of Translational Medicine. 2019 Apr;7(7):152–2.
19.     Emanuel KS, Dahmen J, Sierevelt IN, Mats Brittberg, Kerkhoffs GMMJ. Regression to the Mean: Statistical Bias Can Mislead Interpretation in Cartilage and Osteoarthritis Clinics and Research. Cartilage. 2024 Nov 11.
20.     Suwanraksa C, Sukcharoen W, Phuakphiuwong S, Kooptisirirat S, Krutsuwan K, Nantasri C, et al. Time series analysis of dosimetric changes in target volumes and organs at risk monitored by cone beam computed tomography during radiotherapy for non-small-cell lung cancer. Physics and Imaging in Radiation Oncology. 2025 Jul;35:100822.
21.     Wang YW, Chen M, Shen WT, Xu HP. The clinical practice and dosimetric outcome of the manual adaptive planning during definitive radiotherapy for cervical cancer. Journal of Cancer Research and Clinical Oncology. 2024 May 27;150(5):280–0.
22.     Hanieh Alimiri Dehbaghi, Karim Khoshgard. Revolutionizing emergency care: an overview of the transformative role of artificial intelligence in diagnosis, triage, and patient management. International Journal of Emergency Medicine. 2025 Nov 17;18(1):242–2.
23.     Luijken K, Song J, Groenwold RHH. Quantitative prediction error analysis to investigate predictive performance under predictor measurement heterogeneity at model implementation. Diagn Progn Res. 2022 Apr 7 [cited 2024 Feb 24];6(1):8.
24.     Liu Y, Tian Y, Wang S, Zhang X, Kwong S. Overview of High-Dynamic-Range Image Quality Assessment. Journal of Imaging. 2024 Sep 27;10(10):243.
 
 
 
 
 
 
Iranian Journal of Medical Physics
Volume 23, Issue 2 - Serial Number 2
March and April 2026
Pages 111-116

  • Receive Date 28 April 2026
  • Revise Date 30 May 2026
  • Accept Date 04 July 2026