Deep Neural Network for Differentiation of Brain Tumor Tissue Displayed by Confocal Laser Endomicroscopy.
brain tumor
confocal laser endomicroscopy
deep neural network
fluorescein sodium
image analysis
machine learning
Journal
Frontiers in oncology
ISSN: 2234-943X
Titre abrégé: Front Oncol
Pays: Switzerland
ID NLM: 101568867
Informations de publication
Date de publication:
2021
2021
Historique:
received:
15
02
2021
accepted:
09
04
2021
entrez:
28
5
2021
pubmed:
29
5
2021
medline:
29
5
2021
Statut:
epublish
Résumé
Reliable on site classification of resected tumor specimens remains a challenge. Implementation of high-resolution confocal laser endoscopic techniques (CLEs) during fluorescence-guided brain tumor surgery is a new tool for intraoperative tumor tissue visualization. To overcome observer dependent errors, we aimed to predict tumor type by applying a deep learning model to image data obtained by CLE. Human brain tumor specimens from 25 patients with brain metastasis, glioblastoma, and meningioma were evaluated within this study. In addition to routine histopathological analysis, tissue samples were stained with fluorescein Multiple CLE images were obtained from each specimen with a total number of 13,972 fluorescein based images. Test accuracy of 90.9% was achieved after applying a two-class prediction for glioblastomas and brain metastases with an area under the curve (AUC) value of 0.92. For three class predictions, our model achieved a ratio of correct predicted label of 85.8% in the test set, which was confirmed with five-fold cross validation, without definition of confidence. Applying a confidence rate of 0.999 increased the prediction accuracy to 98.6% when images with substantial artifacts were excluded before the analysis. 36.3% of total images met the output criteria. We trained a residual network model that allows automated, on site analysis of resected tumor specimens based on CLE image datasets. Further
Sections du résumé
BACKGROUND
BACKGROUND
Reliable on site classification of resected tumor specimens remains a challenge. Implementation of high-resolution confocal laser endoscopic techniques (CLEs) during fluorescence-guided brain tumor surgery is a new tool for intraoperative tumor tissue visualization. To overcome observer dependent errors, we aimed to predict tumor type by applying a deep learning model to image data obtained by CLE.
METHODS
METHODS
Human brain tumor specimens from 25 patients with brain metastasis, glioblastoma, and meningioma were evaluated within this study. In addition to routine histopathological analysis, tissue samples were stained with fluorescein
RESULTS
RESULTS
Multiple CLE images were obtained from each specimen with a total number of 13,972 fluorescein based images. Test accuracy of 90.9% was achieved after applying a two-class prediction for glioblastomas and brain metastases with an area under the curve (AUC) value of 0.92. For three class predictions, our model achieved a ratio of correct predicted label of 85.8% in the test set, which was confirmed with five-fold cross validation, without definition of confidence. Applying a confidence rate of 0.999 increased the prediction accuracy to 98.6% when images with substantial artifacts were excluded before the analysis. 36.3% of total images met the output criteria.
CONCLUSIONS
CONCLUSIONS
We trained a residual network model that allows automated, on site analysis of resected tumor specimens based on CLE image datasets. Further
Identifiants
pubmed: 34046358
doi: 10.3389/fonc.2021.668273
pmc: PMC8147727
doi:
Types de publication
Journal Article
Langues
eng
Pagination
668273Informations de copyright
Copyright © 2021 Ziebart, Stadniczuk, Roos, Ratliff, von Deimling, Hänggi and Enders.
Déclaration de conflit d'intérêts
DS was employed by the company Clevertech Inc. The remaining 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
J Cytol. 2016 Apr-Jun;33(2):93-7
pubmed: 27279685
Int J Comput Assist Radiol Surg. 2019 Jan;14(1):31-42
pubmed: 30078151
Radiology. 2020 May;295(2):407-415
pubmed: 32181729
Stem Cell Res Ther. 2019 May 27;10(1):148
pubmed: 31133056
Head Neck. 2016 Apr;38 Suppl 1:E1419-26
pubmed: 26560348
United European Gastroenterol J. 2015 Jun;3(3):230-54
pubmed: 26137298
Clin Neurol Neurosurg. 2018 Jun;169:21-28
pubmed: 29604507
Surg Innov. 2019 Apr;26(2):141-148
pubmed: 30466375
J Med Imaging (Bellingham). 2019 Jan;6(1):014502
pubmed: 30840732
J Neurosurg Sci. 2018 Dec;62(6):704-717
pubmed: 30160080
Neurosurg Focus. 2016 Mar;40(3):E11
pubmed: 26926051
Rinsho Byori. 1992 Dec;40(12):1319-28
pubmed: 1307638
Gastrointest Endosc. 2019 Mar;89(3):626-636
pubmed: 30120955
Biomed Res Int. 2016;2016:6183218
pubmed: 27127791
Nature. 2015 May 28;521(7553):436-44
pubmed: 26017442
Front Oncol. 2019 Jul 03;9:554
pubmed: 31334106
IEEE J Biomed Health Inform. 2020 Oct;24(10):2883-2893
pubmed: 32203040
Acta Otorhinolaryngol Ital. 2019 Dec;39(6):389-395
pubmed: 30745593
Sci Rep. 2017 Sep 20;7(1):11979
pubmed: 28931888
Sci Rep. 2019 Dec 27;9(1):20010
pubmed: 31882817
Front Oncol. 2020 Dec 04;10:599250
pubmed: 33344251
World Neurosurg. 2017 Apr;100:180-185
pubmed: 28069420
Ann Surg Oncol. 2019 Apr;26(4):1158-1165
pubmed: 30719635
Neurosurg Focus. 2014 Feb;36(2):E16
pubmed: 24484254
Microcirculation. 2021 Apr;28(3):e12678
pubmed: 33426724
Neurosurgery. 2011 Jun;68(2 Suppl Operative):282-90; discussion 290
pubmed: 21336204
J Neurosurg. 2018 Sep 7;131(3):724-734
pubmed: 30192200
Respiration. 2019;97(3):259-263
pubmed: 30428462
Front Oncol. 2019 Jun 25;9:519
pubmed: 31293966
Nat Med. 2020 Jan;26(1):52-58
pubmed: 31907460
Surg Neurol Int. 2014 Dec 03;5:170
pubmed: 25593754
Front Oncol. 2019 Jul 18;9:583
pubmed: 31380264