Robustness of radiomic features in magnetic resonance imaging for patients with glioblastoma: Multi-center study.
Features stability
Glioblastoma multiforme
Image normalization
Prognostic modelling
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 2022
Apr 2022
Historique:
received:
10
10
2021
revised:
06
05
2022
accepted:
11
05
2022
entrez:
31
5
2022
pubmed:
1
6
2022
medline:
1
6
2022
Statut:
epublish
Résumé
Radiomics offers great potential in improving diagnosis and treatment for patients with glioblastoma multiforme. However, in order to implement radiomics in clinical routine, the features used for prognostic modelling need to be stable. This comprises significant challenge in multi-center studies. The aim of this study was to evaluate the impact of different image normalization methods on MRI features robustness in multi-center study. Radiomics stability was checked on magnetic resonance images of eleven patients. The images were acquired in two different hospitals using contrast-enhanced T1 sequences. The images were normalized using one of five investigated approaches including grey-level discretization, histogram matching and z-score. Then, radiomic features were extracted and features stability was evaluated using intra-class correlation coefficients. In the second part of the study, improvement in the prognostic performance of features was tested on 60 patients derived from publicly available dataset. Depending on the normalization scheme, the percentage of stable features varied from 3.4% to 8%. The histogram matching based on the tumor region showed the highest amount of the stable features (113/1404); while normalization using fixed bin size resulted in 48 stable features. The histogram matching also led to better prognostic value (median c-index increase of 0.065) comparing to non-normalized images. MRI normalization plays an important role in radiomics. Appropriate normalization helps to select robust features, which can be used for prognostic modelling in multicenter studies. In our study, histogram matching based on tumor region improved both stability of radiomic features and their prognostic value.
Sections du résumé
Background and purpose
UNASSIGNED
Radiomics offers great potential in improving diagnosis and treatment for patients with glioblastoma multiforme. However, in order to implement radiomics in clinical routine, the features used for prognostic modelling need to be stable. This comprises significant challenge in multi-center studies. The aim of this study was to evaluate the impact of different image normalization methods on MRI features robustness in multi-center study.
Methods
UNASSIGNED
Radiomics stability was checked on magnetic resonance images of eleven patients. The images were acquired in two different hospitals using contrast-enhanced T1 sequences. The images were normalized using one of five investigated approaches including grey-level discretization, histogram matching and z-score. Then, radiomic features were extracted and features stability was evaluated using intra-class correlation coefficients. In the second part of the study, improvement in the prognostic performance of features was tested on 60 patients derived from publicly available dataset.
Results
UNASSIGNED
Depending on the normalization scheme, the percentage of stable features varied from 3.4% to 8%. The histogram matching based on the tumor region showed the highest amount of the stable features (113/1404); while normalization using fixed bin size resulted in 48 stable features. The histogram matching also led to better prognostic value (median c-index increase of 0.065) comparing to non-normalized images.
Conclusions
UNASSIGNED
MRI normalization plays an important role in radiomics. Appropriate normalization helps to select robust features, which can be used for prognostic modelling in multicenter studies. In our study, histogram matching based on tumor region improved both stability of radiomic features and their prognostic value.
Identifiants
pubmed: 35633866
doi: 10.1016/j.phro.2022.05.006
pii: S2405-6316(22)00046-X
pmc: PMC9130546
doi:
Types de publication
Journal Article
Langues
eng
Pagination
131-136Commentaires et corrections
Type : ErratumIn
Informations de copyright
© 2022 The Authors. Published by Elsevier B.V. on behalf of European Society of Radiotherapy & Oncology.
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
Clin Neurol Neurosurg. 2016 Dec;151:31-36
pubmed: 27764705
Med Phys. 2020 Sep;47(9):4265-4280
pubmed: 32615647
Eur Radiol. 2021 Aug;31(8):6001-6012
pubmed: 33492473
Front Oncol. 2019 May 21;9:374
pubmed: 31165039
PLoS One. 2017 Sep 21;12(9):e0178524
pubmed: 28934225
Med Phys. 2009 Apr;36(4):1236-43
pubmed: 19472631
Stat Med. 1998 Jan 15;17(1):101-10
pubmed: 9463853
Sci Rep. 2020 Jul 23;10(1):12340
pubmed: 32704007
Hum Brain Mapp. 2019 Dec 1;40(17):4952-4964
pubmed: 31403237
Q J Nucl Med Mol Imaging. 2019 Dec;63(4):355-370
pubmed: 31527578
Lancet Oncol. 2014 Aug;15(9):e395-403
pubmed: 25079102
Invest Radiol. 2017 Jun;52(6):360-366
pubmed: 28079702
Invest Radiol. 2019 Apr;54(4):221-228
pubmed: 30433891
AJNR Am J Neuroradiol. 2018 Aug;39(8):1439-1445
pubmed: 30002055
Crit Rev Oncol Hematol. 2021 Jan;157:103188
pubmed: 33307200
Sci Rep. 2019 Dec 19;9(1):19411
pubmed: 31857632
BMC Med Imaging. 2021 Feb 3;21(1):17
pubmed: 33535988
IEEE J Biomed Health Inform. 2019 Mar;23(2):795-804
pubmed: 29993848
Sci Rep. 2020 Sep 23;10(1):15496
pubmed: 32968131
Cancer Imaging. 2020 Aug 5;20(1):55
pubmed: 32758279
Magn Reson Med. 2009 Dec;62(6):1609-18
pubmed: 19859947
J Digit Imaging. 2014 Dec;27(6):805-23
pubmed: 24990346
Sci Data. 2017 Sep 05;4:170117
pubmed: 28872634
J Neurosurg. 2014 Nov;121(5):1115-23
pubmed: 25192475
Eur J Cancer. 2012 Mar;48(4):441-6
pubmed: 22257792
Front Oncol. 2021 Apr 14;11:636672
pubmed: 33937035
Ann ICRP. 1989;20(4):1-57
pubmed: 2699178
Sci Rep. 2019 Mar 18;9(1):4800
pubmed: 30886309
PLoS One. 2019 Mar 7;14(3):e0213459
pubmed: 30845221
Biometrics. 1989 Mar;45(1):255-68
pubmed: 2720055
Eur Radiol. 2021 Apr;31(4):2272-2280
pubmed: 32975661