Design and validation of a deep learning GAN for predicting 1 7 7 Lu dose voxel kernels using GATE/GEANT4 Monte Carlo and IDAC‐Dose 2.1 in phantom dosimetry
Journal of Applied Clinical Medical Physics · 2026-06 · 27 (6)
DOI: 10.1002/acm2.70661 ↗Kurz gefasst
Validierungsstudie in der Zeitschrift Journal of Applied Clinical Medical Physics, online erschienen am 15.06.2026. Verfasst von Otieno Erick Kapis und 3 weiteren Autor:innen. Beteiligte Einrichtung: Department of Physics, University of Nairobi, Nairobi, Kenya. Hauptthemen laut MEDLINE: Monte Carlo Method, Phantoms, Imaging, Radiotherapy Planning, Computer-Assisted, Radiometry. Die Arbeit nennt 42 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Background
The rapid evolution of deep learning (DL), particularly the emergence of generative adversarial networks (GANs) has demonstrated strong potential for generating high‐resolution, voxel‐level dosimetric data. However, GAN‐based approaches have not yet been broadly applied to the generation of tissue‐specific dose voxel kernels (DVKs) in internal dosimetry. This gap underscores a critical unmet need for methods that can deliver Monte Carlo–level accuracy while achieving substantially improved computational efficiency.
Purpose
This study aimed to design a GAN architecture to train models that can generate 177 Lu DVKs for internal radiation dosimetry comparable to Monte Carlo standards. Materials and methods A GAN model was trained with paired CT‐derived density kernel maps (DKs; voxel size 0.24 3 cm 3 , matrix 15 × 15 × 15) and their corresponding Monte Carlo‐simulated 177 Lu DVKs. The GAN‐generated DVKs were applied to SPECT images of spherical phantoms with varying volumes and initial 177 Lu activity concentrations to compute mean absorbed doses. Results were validated against full GATE Monte Carlo simulations and IDAC‐Dose 2.1 application.
Results
The GAN model predicted 177 Lu DVKs for water, soft tissue, muscle, and cortical bone with accuracies of 98.2%, 98.3%, 96.9%, and 94.2%, respectively ( p > 0.05). Comparisons of GAN‐derived 177 Lu DVKs with both GATE Monte Carlo and IDAC‐Dose 2.1 produced very strong correlations (Pearson's r > 0.99; p = 0.0001). Bland–Altman analysis showed narrow limits of agreement at the 95% confidence interval (CI: −0.59 Gy to 0.44 Gy for GATE MC vs. GAN DVK, and CI: −0.01 Gy to −0.21 Gy for IDAC‐Dose 2.1 vs. GAN DVK), with the vast majority of points falling within these ranges. These findings confirm the robustness and reliability of the GAN‐predicted DVKs when benchmarked against established dosimetric methods.
Conclusion
GAN‐generated 177 Lu DVK convolution provides a potentially fast and computationally efficient approach for internal dosimetry, with strong agreement to established methods in phantom studies. While the clinical impact of tissue‐specific DVKs remains dependent on broader improvements in quantitative imaging and dose–response characterization, this approach represents a promising proof‐of‐concept toward more physically consistent and scalable voxel‐level dosimetry.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Substanzen
Autor:innen und Institutionen
- Department of Physics, University of Nairobi, Nairobi, Kenya.
- Department of Radiation Oncology, Cancer Treatment Center, The Nairobi Hospital, Nairobi, Kenya.
- Laboratory of Medical Physics, IRCCS Istituto Nazionale Tumori Regina Elena, Rome, Italy.
Zitieren
APA
Kapis, O. E., Kaniu, M. I., Giuseppe, I., & Angeyo, H. K. (2026). Design and validation of a deep learning GAN for predicting 1 7 7 Lu dose voxel kernels using GATE/GEANT4 Monte Carlo and IDAC‐Dose 2.1 in phantom dosimetry. Journal of Applied Clinical Medical Physics, 27(6). https://doi.org/10.1002/acm2.70661
Vancouver
Kapis OE, Kaniu MI, Giuseppe I, Angeyo HK. Design and validation of a deep learning GAN for predicting 1 7 7 Lu dose voxel kernels using GATE/GEANT4 Monte Carlo and IDAC‐Dose 2.1 in phantom dosimetry. Journal of Applied Clinical Medical Physics. 2026;27(6). doi:10.1002/acm2.70661
BibTeX
@article{Kapis2026,
author = {Kapis, Otieno Erick and Kaniu, Muchai Ian and Giuseppe, Iaccarino and Angeyo, Hudson Kalambuka},
title = {Design and validation of a deep learning GAN for predicting 1 7 7 Lu dose voxel kernels using GATE/GEANT4 Monte Carlo and IDAC‐Dose 2.1 in phantom dosimetry},
journal = {Journal of Applied Clinical Medical Physics},
year = {2026},
volume = {27},
number = {6},
doi = {10.1002/acm2.70661}
}
EUVIMED