Machine learning decision tree models for multiclass classification of common malignant brain tumors using perfusion and spectroscopy MRI data.
classification and regression tree (CART)
glioblastoma
lymphoma
metastasis
multiclass classification
Journal
Frontiers in oncology
ISSN: 2234-943X
Titre abrégé: Front Oncol
Pays: Switzerland
ID NLM: 101568867
Informations de publication
Date de publication:
2023
2023
Historique:
received:
08
11
2022
accepted:
17
07
2023
medline:
24
8
2023
pubmed:
24
8
2023
entrez:
24
8
2023
Statut:
epublish
Résumé
To investigate the contribution of machine learning decision tree models applied to perfusion and spectroscopy MRI for multiclass classification of lymphomas, glioblastomas, and metastases, and then to bring out the underlying key pathophysiological processes involved in the hierarchization of the decision-making algorithms of the models. From 2013 to 2020, 180 consecutive patients with histopathologically proved lymphomas (n = 77), glioblastomas (n = 45), and metastases (n = 58) were included in machine learning analysis after undergoing MRI. The perfusion parameters (rCBV The decision tree model thus constructed successfully classified all 3 tumor types with a performance (AUC) of 0.98 for PCNSLs, 0.98 for GBM and 1.00 for METs. The model accuracy was 0.96 with a RSquare of 0.887. Five rules of classifier combinations were extracted with a predicted probability from 0.907 to 0.989 for that end nodes of the decision tree for tumor multiclass classification. In hierarchical order of importance, the root node (Cho/NAA) in the decision tree algorithm was primarily based on the proliferative, infiltrative, and neuronal destructive characteristics of the tumor, the internal node (PSRmax), on tumor tissue capillary permeability characteristics, and the end node (Lac/Cr or Cho/Cr), on tumor energy glycolytic (Warburg effect), or on membrane lipid tumor metabolism. Our study shows potential implementation of machine learning decision tree model algorithms based on a hierarchical, convenient, and personalized use of perfusion and spectroscopy MRI data for multiclass classification of these brain tumors.
Sections du résumé
Background
UNASSIGNED
To investigate the contribution of machine learning decision tree models applied to perfusion and spectroscopy MRI for multiclass classification of lymphomas, glioblastomas, and metastases, and then to bring out the underlying key pathophysiological processes involved in the hierarchization of the decision-making algorithms of the models.
Methods
UNASSIGNED
From 2013 to 2020, 180 consecutive patients with histopathologically proved lymphomas (n = 77), glioblastomas (n = 45), and metastases (n = 58) were included in machine learning analysis after undergoing MRI. The perfusion parameters (rCBV
Results
UNASSIGNED
The decision tree model thus constructed successfully classified all 3 tumor types with a performance (AUC) of 0.98 for PCNSLs, 0.98 for GBM and 1.00 for METs. The model accuracy was 0.96 with a RSquare of 0.887. Five rules of classifier combinations were extracted with a predicted probability from 0.907 to 0.989 for that end nodes of the decision tree for tumor multiclass classification. In hierarchical order of importance, the root node (Cho/NAA) in the decision tree algorithm was primarily based on the proliferative, infiltrative, and neuronal destructive characteristics of the tumor, the internal node (PSRmax), on tumor tissue capillary permeability characteristics, and the end node (Lac/Cr or Cho/Cr), on tumor energy glycolytic (Warburg effect), or on membrane lipid tumor metabolism.
Conclusion
UNASSIGNED
Our study shows potential implementation of machine learning decision tree model algorithms based on a hierarchical, convenient, and personalized use of perfusion and spectroscopy MRI data for multiclass classification of these brain tumors.
Identifiants
pubmed: 37614505
doi: 10.3389/fonc.2023.1089998
pmc: PMC10442801
doi:
Types de publication
Journal Article
Langues
eng
Pagination
1089998Informations de copyright
Copyright © 2023 Vallée, Vallée, Guillevin, Lallouette, Thomas, Rittano, Wager, Guillevin and Vallée.
Déclaration de conflit d'intérêts
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Références
Clin Neuroradiol. 2014 Dec;24(4):329-36
pubmed: 23994941
Med Image Comput Comput Assist Interv. 2005;8(Pt 1):506-14
pubmed: 16685884
Neuroradiology. 2003 Jan;45(1):44-9
pubmed: 12525954
Nature. 2015 May 28;521(7553):436-44
pubmed: 26017442
ACS Chem Neurosci. 2020 Feb 5;11(3):477-483
pubmed: 31922391
AJNR Am J Neuroradiol. 1996 Jan;17(1):1-15
pubmed: 8770242
Neuroimaging Clin N Am. 2013 Aug;23(3):359-80
pubmed: 23928194
AJNR Am J Neuroradiol. 2011 Mar;32(3):507-14
pubmed: 21330399
Front Oncol. 2020 Feb 07;10:71
pubmed: 32117728
AJNR Am J Neuroradiol. 1995 Sep;16(8):1593-603
pubmed: 7502961
Neuroradiology. 2003 Dec;45(12):865-8
pubmed: 14605786
Oncotarget. 2017 May 10;8(35):59492-59499
pubmed: 28938652
Comput Inform Nurs. 2016 Apr;34(4):175-82
pubmed: 26848645
PLoS One. 2010 Jul 20;5(7):e11625
pubmed: 20652023
AJNR Am J Neuroradiol. 2005 Oct;26(9):2187-99
pubmed: 16219821
Eur Radiol. 2001;11(9):1784-91
pubmed: 11511902
AJNR Am J Neuroradiol. 2007 Jun-Jul;28(6):1078-84
pubmed: 17569962
AJNR Am J Neuroradiol. 2018 Aug;39(8):1423-1431
pubmed: 30049719
Neurology. 2008 Apr 15;70(16):1353-7
pubmed: 18413589
AJNR Am J Neuroradiol. 2011 Jun-Jul;32(6):1004-10
pubmed: 21511863
Comput Methods Programs Biomed. 2017 Feb;139:83-91
pubmed: 28187897
Eur Radiol. 2021 Nov;31(11):8703-8713
pubmed: 33890149
J Comput Assist Tomogr. 2010 Nov-Dec;34(6):836-41
pubmed: 21084897
Prog Neurobiol. 2007 Feb;81(2):89-131
pubmed: 17275978
Eur Radiol. 2018 Sep;28(9):3819-3831
pubmed: 29619517
Magn Reson Imaging Clin N Am. 2016 Feb;24(1):87-122
pubmed: 26613877
AJR Am J Roentgenol. 2002 Sep;179(3):783-9
pubmed: 12185064
Radiology. 1994 Apr;191(1):41-51
pubmed: 8134596
AJNR Am J Neuroradiol. 2013 Jun-Jul;34(6):1145-9
pubmed: 23348763
BBA Clin. 2016 Apr 12;5:170-8
pubmed: 27158592
J Magn Reson Imaging. 2006 Oct;24(4):817-24
pubmed: 16958061
Neuroimaging Clin N Am. 2010 Aug;20(3):293-310
pubmed: 20708548
Cancer Imaging. 2006 Jun 22;6:95-9
pubmed: 16829470
J Magn Reson Imaging. 2010 Nov;32(5):1038-44
pubmed: 21031506
J Neuroradiol. 2012 Dec;39(5):301-7
pubmed: 22197404
Comput Biol Med. 2021 Dec 4;140:105111
pubmed: 34891095
Acta Radiol. 2010 Apr;51(3):316-25
pubmed: 20092374
Life Sci. 1996;58(22):1929-35
pubmed: 8637421
AJNR Am J Neuroradiol. 2006 Apr;27(4):859-67
pubmed: 16611779
Comput Biol Med. 2019 May;108:354-370
pubmed: 31054502
Top Magn Reson Imaging. 2004 Oct;15(5):291-313
pubmed: 15627004
J Magn Reson. 2005 Apr;173(2):218-28
pubmed: 15780914
Cancers (Basel). 2021 Nov 05;13(21):
pubmed: 34771718
Clin Neurol Neurosurg. 2005 Aug;107(5):379-84
pubmed: 16023531
J Neurooncol. 2008 Jan;86(2):225-9
pubmed: 17786533
Biochim Biophys Acta. 2013 Mar;1831(3):523-32
pubmed: 23010477
Clin Radiol. 2004 Jan;59(1):77-85
pubmed: 14697379
Rev Neurosci. 2018 Jan 26;29(1):71-91
pubmed: 28822229
Artif Intell Med. 1999 May;16(1):3-23
pubmed: 10225344
Neuroradiology. 2011 May;53(5):319-30
pubmed: 20625709
Neural Netw. 2015 Jan;61:85-117
pubmed: 25462637
Radiology. 1992 Jun;183(3):701-9
pubmed: 1584924
Neurosci Lett. 2003 May 22;342(3):163-6
pubmed: 12757890
Radiology. 1993 Mar;186(3):745-52
pubmed: 8430183