Deep learning-based automatic tumor burden assessment of pediatric high-grade gliomas, medulloblastomas, and other leptomeningeal seeding tumors.
brain
deep learning
response assessment
segmentation
Journal
Neuro-oncology
ISSN: 1523-5866
Titre abrégé: Neuro Oncol
Pays: England
ID NLM: 100887420
Informations de publication
Date de publication:
01 02 2022
01 02 2022
Historique:
pubmed:
27
6
2021
medline:
23
3
2022
entrez:
26
6
2021
Statut:
ppublish
Résumé
Longitudinal measurement of tumor burden with magnetic resonance imaging (MRI) is an essential component of response assessment in pediatric brain tumors. We developed a fully automated pipeline for the segmentation of tumors in pediatric high-grade gliomas, medulloblastomas, and leptomeningeal seeding tumors. We further developed an algorithm for automatic 2D and volumetric size measurement of tumors. The preoperative and postoperative cohorts were randomly split into training and testing sets in a 4:1 ratio. A 3D U-Net neural network was trained to automatically segment the tumor on T1 contrast-enhanced and T2/FLAIR images. The product of the maximum bidimensional diameters according to the RAPNO (Response Assessment in Pediatric Neuro-Oncology) criteria (AutoRAPNO) was determined. Performance was compared to that of 2 expert human raters who performed assessments independently. Volumetric measurements of predicted and expert segmentations were computationally derived and compared. A total of 794 preoperative MRIs from 794 patients and 1003 postoperative MRIs from 122 patients were included. There was excellent agreement of volumes between preoperative and postoperative predicted and manual segmentations, with intraclass correlation coefficients (ICCs) of 0.912 and 0.960 for the 2 preoperative and 0.947 and 0.896 for the 2 postoperative models. There was high agreement between AutoRAPNO scores on predicted segmentations and manually calculated scores based on manual segmentations (Rater 2 ICC = 0.909; Rater 3 ICC = 0.851). Lastly, the performance of AutoRAPNO was superior in repeatability to that of human raters for MRIs with multiple lesions. Our automated deep learning pipeline demonstrates potential utility for response assessment in pediatric brain tumors. The tool should be further validated in prospective studies.
Sections du résumé
BACKGROUND
Longitudinal measurement of tumor burden with magnetic resonance imaging (MRI) is an essential component of response assessment in pediatric brain tumors. We developed a fully automated pipeline for the segmentation of tumors in pediatric high-grade gliomas, medulloblastomas, and leptomeningeal seeding tumors. We further developed an algorithm for automatic 2D and volumetric size measurement of tumors.
METHODS
The preoperative and postoperative cohorts were randomly split into training and testing sets in a 4:1 ratio. A 3D U-Net neural network was trained to automatically segment the tumor on T1 contrast-enhanced and T2/FLAIR images. The product of the maximum bidimensional diameters according to the RAPNO (Response Assessment in Pediatric Neuro-Oncology) criteria (AutoRAPNO) was determined. Performance was compared to that of 2 expert human raters who performed assessments independently. Volumetric measurements of predicted and expert segmentations were computationally derived and compared.
RESULTS
A total of 794 preoperative MRIs from 794 patients and 1003 postoperative MRIs from 122 patients were included. There was excellent agreement of volumes between preoperative and postoperative predicted and manual segmentations, with intraclass correlation coefficients (ICCs) of 0.912 and 0.960 for the 2 preoperative and 0.947 and 0.896 for the 2 postoperative models. There was high agreement between AutoRAPNO scores on predicted segmentations and manually calculated scores based on manual segmentations (Rater 2 ICC = 0.909; Rater 3 ICC = 0.851). Lastly, the performance of AutoRAPNO was superior in repeatability to that of human raters for MRIs with multiple lesions.
CONCLUSIONS
Our automated deep learning pipeline demonstrates potential utility for response assessment in pediatric brain tumors. The tool should be further validated in prospective studies.
Identifiants
pubmed: 34174070
pii: 6310123
doi: 10.1093/neuonc/noab151
pmc: PMC8804897
doi:
Types de publication
Journal Article
Research Support, N.I.H., Extramural
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
289-299Subventions
Organisme : NCI NIH HHS
ID : F30 CA239407
Pays : United States
Organisme : NIBIB NIH HHS
ID : T32 EB001680
Pays : United States
Commentaires et corrections
Type : ErratumIn
Type : CommentIn
Informations de copyright
© The Author(s) 2021. Published by Oxford University Press on behalf of the Society for Neuro-Oncology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
Références
J Neurooncol. 1999 May;42(3):271-88
pubmed: 10433110
Neuro Oncol. 2015 Sep;17(9):1188-98
pubmed: 26250565
Phys Med Biol. 2011 Jul 21;56(14):4557-77
pubmed: 21725140
Psychol Rep. 1966 Aug;19(1):3-11
pubmed: 5942109
AJNR Am J Neuroradiol. 2020 Sep;41(9):1718-1725
pubmed: 32816765
J Neurooncol. 2018 Mar;137(1):103-110
pubmed: 29204840
Neurosurgery. 2000 Oct;47(4):879-85; discussion 885-7
pubmed: 11014428
Front Comput Neurosci. 2020 Apr 08;14:25
pubmed: 32322196
Curr Treat Options Neurol. 2013 Jun;15(3):316-27
pubmed: 23440592
Lancet Oncol. 2020 Jun;21(6):e317-e329
pubmed: 32502458
Neuro Oncol. 2019 Nov 4;21(11):1412-1422
pubmed: 31190077
Neurology. 2003 Mar 11;60(5):826-30
pubmed: 12629241
J Neurooncol. 2008 Oct;90(1):57-61
pubmed: 18587536
Neuro Oncol. 2020 Oct 14;22(10):1536-1544
pubmed: 32215549
Med Image Anal. 2017 Feb;36:61-78
pubmed: 27865153
IEEE Trans Med Imaging. 2011 Sep;30(9):1617-34
pubmed: 21880566
Nat Methods. 2020 Mar;17(3):261-272
pubmed: 32015543
Neuroinformatics. 2021 Jan;19(1):127-140
pubmed: 32578020
Psychol Bull. 1979 Mar;86(2):420-8
pubmed: 18839484
Med Image Anal. 2017 Jan;35:18-31
pubmed: 27310171
Dev Cogn Neurosci. 2018 Oct;33:206-223
pubmed: 29033222
IEEE Trans Med Imaging. 2010 Jun;29(6):1310-20
pubmed: 20378467
Neuro Oncol. 2018 Jan 10;20(1):13-23
pubmed: 28449033
Neuro Oncol. 2017 Jun 1;19(6):744-745
pubmed: 28379518
Neuro Oncol. 2006 Jan;8(1):38-46
pubmed: 16443946
Neuro Oncol. 2013 Jul;15(7):945-54
pubmed: 23788270
Neurol Clin. 2018 Aug;36(3):533-556
pubmed: 30072070