Automated detection of dental artifacts for large-scale radiomic analysis in radiation oncology.
Computed tomography
Deep learning
Metal artifact detection
Radiomics
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:
Apr 2021
Apr 2021
Historique:
received:
23
09
2020
revised:
09
02
2021
accepted:
06
04
2021
entrez:
14
7
2021
pubmed:
15
7
2021
medline:
15
7
2021
Statut:
epublish
Résumé
Computed tomography (CT) is one of the most common medical imaging modalities in radiation oncology and radiomics research, the computational voxel-level analysis of medical images. Radiomics is vulnerable to the effects of dental artifacts (DA) caused by metal implants or fillings and can hamper future reproducibility on new datasets. In this study we seek to better understand the robustness of quantitative radiomic features to DAs. Furthermore, we propose a novel method of detecting DAs in order to safeguard radiomic studies and improve reproducibility. We analyzed the correlations between radiomic features and the location of dental artifacts in a new dataset containing 3D CT scans from 3211 patients. We then combined conventional image processing techniques with a pre-trained convolutional neural network to create a three-class patient-level DA classifier and slice-level DA locator. Finally, we demonstrated its utility in reducing the correlations between the location of DAs and certain radiomic features. We found that when strong DAs were present, the proximity of the tumour to the mouth was highly correlated with 36 radiomic features. We predicted the correct DA magnitude yielding a Matthews correlation coefficient of 0.73 and location of DAs achieving the same level of agreement as human labellers. Removing radiomic features or CT slices containing DAs could reduce the unwanted correlations between the location of DAs and radiomic features. Automated DA detection can be used to improve the reproducibility of radiomic studies; an important step towards creating effective radiomic models for use in clinical radiation oncology.
Sections du résumé
BACKGROUND AND PURPOSE
OBJECTIVE
Computed tomography (CT) is one of the most common medical imaging modalities in radiation oncology and radiomics research, the computational voxel-level analysis of medical images. Radiomics is vulnerable to the effects of dental artifacts (DA) caused by metal implants or fillings and can hamper future reproducibility on new datasets. In this study we seek to better understand the robustness of quantitative radiomic features to DAs. Furthermore, we propose a novel method of detecting DAs in order to safeguard radiomic studies and improve reproducibility.
MATERIALS AND METHODS
METHODS
We analyzed the correlations between radiomic features and the location of dental artifacts in a new dataset containing 3D CT scans from 3211 patients. We then combined conventional image processing techniques with a pre-trained convolutional neural network to create a three-class patient-level DA classifier and slice-level DA locator. Finally, we demonstrated its utility in reducing the correlations between the location of DAs and certain radiomic features.
RESULTS
RESULTS
We found that when strong DAs were present, the proximity of the tumour to the mouth was highly correlated with 36 radiomic features. We predicted the correct DA magnitude yielding a Matthews correlation coefficient of 0.73 and location of DAs achieving the same level of agreement as human labellers.
CONCLUSIONS
CONCLUSIONS
Removing radiomic features or CT slices containing DAs could reduce the unwanted correlations between the location of DAs and radiomic features. Automated DA detection can be used to improve the reproducibility of radiomic studies; an important step towards creating effective radiomic models for use in clinical radiation oncology.
Identifiants
pubmed: 34258406
doi: 10.1016/j.phro.2021.04.001
pii: S2405-6316(21)00021-X
pmc: PMC8254196
doi:
Types de publication
Journal Article
Langues
eng
Pagination
41-47Informations de copyright
© 2021 The Author(s).
Déclaration de conflit d'intérêts
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Références
Phys Med Biol. 2020 Jan 10;65(1):015005
pubmed: 31683260
Acta Oncol. 2017 Jun;56(6):874-878
pubmed: 28464749
Phys Imaging Radiat Oncol. 2019 Jun 06;10:49-54
pubmed: 33458268
Comput Med Imaging Graph. 2018 Nov;69:134-139
pubmed: 30268005
Acta Oncol. 2015;54(9):1423-9
pubmed: 26264429
Cancer Res. 2017 Nov 1;77(21):e104-e107
pubmed: 29092951
Radiother Oncol. 2006 May;79(2):198-202
pubmed: 16677729
IEEE Trans Med Imaging. 2018 Jun;37(6):1370-1381
pubmed: 29870366
Med Image Anal. 2017 Dec;42:60-88
pubmed: 28778026
Nucl Med Mol Imaging. 2019 Jun;53(3):164-166
pubmed: 31231435
Nat Commun. 2014 Jun 03;5:4006
pubmed: 24892406
Radiother Oncol. 2019 Jan;130:2-9
pubmed: 30416044