Deep-Learning-Based Dose Predictor for Glioblastoma-Assessing the Sensitivity and Robustness for Dose Awareness in Contouring.
VMAT
deep learning
dose prediction
glioblastoma
quality assurance
radiotherapy
Journal
Cancers
ISSN: 2072-6694
Titre abrégé: Cancers (Basel)
Pays: Switzerland
ID NLM: 101526829
Informations de publication
Date de publication:
23 Aug 2023
23 Aug 2023
Historique:
received:
03
08
2023
revised:
16
08
2023
accepted:
21
08
2023
medline:
9
9
2023
pubmed:
9
9
2023
entrez:
9
9
2023
Statut:
epublish
Résumé
External beam radiation therapy requires a sophisticated and laborious planning procedure. To improve the efficiency and quality of this procedure, machine-learning models that predict these dose distributions were introduced. The most recent dose prediction models are based on deep-learning architectures called 3D U-Nets that give good approximations of the dose in 3D almost instantly. Our purpose was to train such a 3D dose prediction model for glioblastoma VMAT treatment and test its robustness and sensitivity for the purpose of quality assurance of automatic contouring. From a cohort of 125 glioblastoma (GBM) patients, VMAT plans were created according to a clinical protocol. The initial model was trained on a cascaded 3D U-Net. A total of 60 cases were used for training, 15 for validation and 20 for testing. The prediction model was tested for sensitivity to dose changes when subject to realistic contour variations. Additionally, the model was tested for robustness by exposing it to a worst-case test set containing out-of-distribution cases. The initially trained prediction model had a dose score of 0.94 Gy and a mean DVH (dose volume histograms) score for all structures of 1.95 Gy. In terms of sensitivity, the model was able to predict the dose changes that occurred due to the contour variations with a mean error of 1.38 Gy. We obtained a 3D VMAT dose prediction model for GBM with limited data, providing good sensitivity to realistic contour variations. We tested and improved the model's robustness by targeted updates to the training set, making it a useful technique for introducing dose awareness in the contouring evaluation and quality assurance process.
Identifiants
pubmed: 37686501
pii: cancers15174226
doi: 10.3390/cancers15174226
pmc: PMC10486555
pii:
doi:
Types de publication
Journal Article
Langues
eng
Subventions
Organisme : Innosuisse - Swiss Innovation Agency
ID : 31274.1 IP-LS
Références
Comput Methods Programs Biomed. 2023 Apr;231:107374
pubmed: 36738608
Med Phys. 2021 Sep;48(9):5574-5582
pubmed: 34101852
Phys Med Biol. 2011 Jul 21;56(14):4557-77
pubmed: 21725140
Pract Radiat Oncol. 2023 May-Jun;13(3):e282-e291
pubmed: 36697347
Med Phys. 2019 Jan;46(1):56-64
pubmed: 30367492
Radiother Oncol. 2023 Jul;184:109663
pubmed: 37059335
Nat Rev Clin Oncol. 2020 Dec;17(12):771-781
pubmed: 32843739
Front Oncol. 2023 Feb 28;13:1041769
pubmed: 36925918
IEEE Trans Med Imaging. 2016 Apr;35(4):1000-12
pubmed: 26660888
Sci Rep. 2019 Jan 31;9(1):1076
pubmed: 30705354
Phys Med Biol. 2018 Dec 04;63(23):235022
pubmed: 30511663
Radiother Oncol. 2022 May;170:198-204
pubmed: 35351537
Med Phys. 2015 Feb;42(2):1048-59
pubmed: 25652517
Radiat Oncol. 2021 Aug 17;16(1):154
pubmed: 34404441
Radiat Oncol. 2015 Nov 19;10:234
pubmed: 26584574
Med Phys. 2012 Dec;39(12):7446-61
pubmed: 23231294
Phys Med Biol. 2020 Apr 08;65(7):075013
pubmed: 32053803
Front Oncol. 2021 Nov 11;11:752007
pubmed: 34858825
Med Phys. 2016 Jan;43(1):378
pubmed: 26745931
Phys Med Biol. 2015 Jul 07;60(13):5199-209
pubmed: 26083863
Med Image Anal. 2021 Oct;73:102161
pubmed: 34293536
Med Phys. 2021 Sep;48(9):5549-5561
pubmed: 34156719
Med Phys. 2018 Jun;45(6):2560-2564
pubmed: 29663417
Radiat Oncol. 2022 Oct 22;17(1):170
pubmed: 36273161
Med Phys. 2022 Mar;49(3):1344-1356
pubmed: 35043971
Front Neuroimaging. 2022 Oct 28;1:1012639
pubmed: 37555149
Phys Imaging Radiat Oncol. 2021 Jan 30;17:65-70
pubmed: 33898781
Phys Med Biol. 2017 Jan 21;62(2):415-431
pubmed: 27997376
Med Phys. 2006 Nov;33(11):4130-48
pubmed: 17153392
Med Phys. 2023 Feb;50(2):1149-1161
pubmed: 36434793
Insights Imaging. 2022 Aug 17;13(1):137
pubmed: 35976491