External validation of deep learning-based contouring of head and neck organs at risk.
Auto-contouring
Contour comparison
Deep learning
Head & neck cancer
Journal
Physics and imaging in radiation oncology
ISSN: 2405-6316
Titre abrégé: Phys Imaging Radiat Oncol
Pays: Netherlands
ID NLM: 101704276
Informations de publication
Date de publication:
Jul 2020
Jul 2020
Historique:
received:
29
02
2020
revised:
29
05
2020
accepted:
27
06
2020
entrez:
18
1
2021
pubmed:
19
1
2021
medline:
19
1
2021
Statut:
epublish
Résumé
Head and neck (HN) radiotherapy can benefit from automatic delineation of tumor and surrounding organs because of the complex anatomy and the regular need for adaptation. The aim of this study was to assess the performance of a commercially available deep learning contouring (DLC) model on an external validation set. The CT-based DLC model, trained at the University Medical Center Groningen (UMCG), was applied to an independent set of 58 patients from the Radboud University Medical Center (RUMC). DLC results were compared to the RUMC manual reference using the Dice similarity coefficient (DSC) and 95th percentile of Hausdorff distance (HD95). Craniocaudal spatial information was added by calculating binned measures. In addition, a qualitative evaluation compared the acceptance of manual and DLC contours in both groups of observers. Good correspondence was shown for the mandible (DSC 0.90; HD95 3.6 mm). Performance was reasonable for the glandular OARs, brainstem and oral cavity (DSC 0.78-0.85, HD95 3.7-7.3 mm). The other aerodigestive tract OARs showed only moderate agreement (DSC 0.53-0.65, HD95 around 9 mm). The binned measures displayed the largest deviations caudally and/or cranially. This study demonstrates that the DLC model can provide a reasonable starting point for delineation when applied to an independent patient cohort. The qualitative evaluation did not reveal large differences in the interpretation of contouring guidelines between RUMC and UMCG observers.
Sections du résumé
BACKGROUND AND PURPOSE
OBJECTIVE
Head and neck (HN) radiotherapy can benefit from automatic delineation of tumor and surrounding organs because of the complex anatomy and the regular need for adaptation. The aim of this study was to assess the performance of a commercially available deep learning contouring (DLC) model on an external validation set.
MATERIALS AND METHODS
METHODS
The CT-based DLC model, trained at the University Medical Center Groningen (UMCG), was applied to an independent set of 58 patients from the Radboud University Medical Center (RUMC). DLC results were compared to the RUMC manual reference using the Dice similarity coefficient (DSC) and 95th percentile of Hausdorff distance (HD95). Craniocaudal spatial information was added by calculating binned measures. In addition, a qualitative evaluation compared the acceptance of manual and DLC contours in both groups of observers.
RESULTS
RESULTS
Good correspondence was shown for the mandible (DSC 0.90; HD95 3.6 mm). Performance was reasonable for the glandular OARs, brainstem and oral cavity (DSC 0.78-0.85, HD95 3.7-7.3 mm). The other aerodigestive tract OARs showed only moderate agreement (DSC 0.53-0.65, HD95 around 9 mm). The binned measures displayed the largest deviations caudally and/or cranially.
CONCLUSIONS
CONCLUSIONS
This study demonstrates that the DLC model can provide a reasonable starting point for delineation when applied to an independent patient cohort. The qualitative evaluation did not reveal large differences in the interpretation of contouring guidelines between RUMC and UMCG observers.
Identifiants
pubmed: 33458320
doi: 10.1016/j.phro.2020.06.006
pii: S2405-6316(20)30031-2
pmc: PMC7807543
doi:
Types de publication
Journal Article
Langues
eng
Pagination
8-15Informations de copyright
© 2020 The Author(s).
Déclaration de conflit d'intérêts
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: University Medical Center Groningen has a research collaboration with Mirada Medical Ltd., Oxford, UK. Mirada Medical Ltd. has provided Radboud University Medical Center with the software for the external validation. Author MJG is an employee of Mirada Medical Ltd.
Références
Int J Radiat Oncol Biol Phys. 2019 Jul 1;104(3):677-684
pubmed: 30836167
Med Phys. 2014 May;41(5):050902
pubmed: 24784366
Med Phys. 2017 Feb;44(2):547-557
pubmed: 28205307
Radiother Oncol. 2018 Feb;126(2):312-317
pubmed: 29208513
Acta Oncol. 2013 Oct;52(7):1417-22
pubmed: 23957565
IEEE Trans Med Imaging. 2015 Oct;34(10):1993-2024
pubmed: 25494501
BMC Cancer. 2017 Mar 21;17(1):208
pubmed: 28327089
Phys Med Biol. 2018 Jul 11;63(14):145007
pubmed: 29882749
Radiother Oncol. 2020 Jan;142:115-123
pubmed: 31653573
Radiother Oncol. 2019 Sep;138:68-74
pubmed: 31146073
Radiother Oncol. 2015 Oct;117(1):83-90
pubmed: 26277855
Int J Radiat Oncol Biol Phys. 2010 Mar 1;76(3 Suppl):S58-63
pubmed: 20171519
IEEE Trans Med Imaging. 2019 Nov;38(11):2654-2664
pubmed: 30969918
Med Phys. 2017 Dec;44(12):6377-6389
pubmed: 28963779
Med Phys. 2018 Nov;45(11):5105-5115
pubmed: 30229951
Radiother Oncol. 2014 Sep;112(3):317-20
pubmed: 25315862
Phys Med Biol. 2017 Nov 10;62(23):8943-8958
pubmed: 28994665
IEEE Trans Med Imaging. 2016 May;35(5):1299-1312
pubmed: 26978662
IEEE Trans Med Imaging. 2015 May;34(5):1018-30
pubmed: 25376036
Laryngoscope. 2020 Jan;130(1):146-153
pubmed: 30756394
Int J Radiat Oncol Biol Phys. 2010 Dec 1;78(5):1356-65
pubmed: 20646872
Comput Biol Med. 2018 Jul 1;98:126-146
pubmed: 29787940
Eur Radiol. 2019 Apr;29(4):1961-1967
pubmed: 30302589
Med Phys. 2018 Oct;45(10):4568-4581
pubmed: 30144101
Front Oncol. 2019 Apr 09;9:239
pubmed: 31024843