End to end stroke triage using cerebrovascular morphology and machine learning.
CNN—convolutional neural network
cerebrovascular disease
collateral circulation
machine learning
segmentation (image processing)
stroke
stroke outcome
Journal
Frontiers in neurology
ISSN: 1664-2295
Titre abrégé: Front Neurol
Pays: Switzerland
ID NLM: 101546899
Informations de publication
Date de publication:
2023
2023
Historique:
received:
30
05
2023
accepted:
20
09
2023
medline:
9
11
2023
pubmed:
9
11
2023
entrez:
9
11
2023
Statut:
epublish
Résumé
Rapid and accurate triage of acute ischemic stroke (AIS) is essential for early revascularization and improved patient outcomes. Response to acute reperfusion therapies varies significantly based on patient-specific cerebrovascular anatomy that governs cerebral blood flow. We present an end-to-end machine learning approach for automatic stroke triage. Employing a validated convolutional neural network (CNN) segmentation model for image processing, we extract each patient's cerebrovasculature and its morphological features from baseline non-invasive angiography scans. These features are used to detect occlusion's presence and the site automatically, and for the first time, to estimate collateral circulation without manual intervention. We then use the extracted cerebrovascular features along with commonly used clinical and imaging parameters to predict the 90 days functional outcome for each patient. The CNN model achieved a segmentation accuracy of 94% based on the Dice similarity coefficient (DSC). The automatic stroke detection algorithm had a sensitivity and specificity of 92% and 94%, respectively. The models for occlusion site detection and automatic collateral grading reached 96% and 87.2% accuracy, respectively. Incorporating the automatically extracted cerebrovascular features significantly improved the 90 days outcome prediction accuracy from 0.63 to 0.83. The fast, automatic, and comprehensive model presented here can improve stroke diagnosis, aid collateral assessment, and enhance prognostication for treatment decisions, using cerebrovascular morphology.
Sections du résumé
Background
UNASSIGNED
Rapid and accurate triage of acute ischemic stroke (AIS) is essential for early revascularization and improved patient outcomes. Response to acute reperfusion therapies varies significantly based on patient-specific cerebrovascular anatomy that governs cerebral blood flow. We present an end-to-end machine learning approach for automatic stroke triage.
Methods
UNASSIGNED
Employing a validated convolutional neural network (CNN) segmentation model for image processing, we extract each patient's cerebrovasculature and its morphological features from baseline non-invasive angiography scans. These features are used to detect occlusion's presence and the site automatically, and for the first time, to estimate collateral circulation without manual intervention. We then use the extracted cerebrovascular features along with commonly used clinical and imaging parameters to predict the 90 days functional outcome for each patient.
Results
UNASSIGNED
The CNN model achieved a segmentation accuracy of 94% based on the Dice similarity coefficient (DSC). The automatic stroke detection algorithm had a sensitivity and specificity of 92% and 94%, respectively. The models for occlusion site detection and automatic collateral grading reached 96% and 87.2% accuracy, respectively. Incorporating the automatically extracted cerebrovascular features significantly improved the 90 days outcome prediction accuracy from 0.63 to 0.83.
Conclusion
UNASSIGNED
The fast, automatic, and comprehensive model presented here can improve stroke diagnosis, aid collateral assessment, and enhance prognostication for treatment decisions, using cerebrovascular morphology.
Identifiants
pubmed: 37941573
doi: 10.3389/fneur.2023.1217796
pmc: PMC10628321
doi:
Types de publication
Journal Article
Langues
eng
Pagination
1217796Informations de copyright
Copyright © 2023 Deshpande, Elliott, Jiang, Tahsili-Fahadan, Kidwell, Wintermark and Laksari.
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. The reviewer NT declared a shared affiliation with the authors AD, JE, CK, and KL at the time of review.
Références
Stroke. 2007 Mar;38(3):1091-6
pubmed: 17272767
Artif Intell Rev. 2022;55(7):5845-5889
pubmed: 35250146
PLoS One. 2014 Feb 10;9(2):e88225
pubmed: 24520356
Oncotarget. 2017 Sep 19;8(46):81529-81537
pubmed: 29113412
Int J Stroke. 2017 Oct;12(8):896-905
pubmed: 28946832
J Biomech. 2007;40(8):1794-805
pubmed: 17045276
Ann Neurol. 2011 Jun;69(6):963-74
pubmed: 21437932
Neurobiol Aging. 2010 Feb;31(2):290-300
pubmed: 18471935
Ann Biomed Eng. 2018 Aug;46(8):1128-1145
pubmed: 29691787
J Cereb Blood Flow Metab. 2015 Sep;35(9):1397-405
pubmed: 26154867
Med Image Anal. 2009 Dec;13(6):819-45
pubmed: 19818675
Brain. 2009 Aug;132(Pt 8):2231-8
pubmed: 19509116
Int J Mol Sci. 2017 Dec 09;18(12):
pubmed: 29232823
Front Neurol. 2018 Nov 08;9:945
pubmed: 30467491
Neurosurgery. 2019 Sep 1;85(3):350-358
pubmed: 30010973
N Engl J Med. 2018 Feb 22;378(8):708-718
pubmed: 29364767
AJR Am J Roentgenol. 2019 Jan;212(1):44-51
pubmed: 30354266
Front Neurol. 2020 Aug 25;11:889
pubmed: 32982920
Stroke. 2017 Jun;48(6):1574-1579
pubmed: 28487329
J Neuroimaging. 2020 Sep;30(5):562-571
pubmed: 32037629
Int J Stroke. 2015 Oct;10(7):1081-6
pubmed: 26061533
Prog Neurobiol. 2014 Apr;115:138-56
pubmed: 24291532
Swiss Med Wkly. 2017 Dec 12;147:w14538
pubmed: 29231236
Cerebrovasc Dis. 2019;47(5-6):217-222
pubmed: 31216543
AJNR Am J Neuroradiol. 2021 Jan;42(2):240-246
pubmed: 33414230
Front Neurol. 2021 May 06;12:663899
pubmed: 34025567
Stroke. 2018 Feb;49(2):498-500
pubmed: 29242390
Sci Rep. 2016 Feb 10;6:20932
pubmed: 26860196
AJNR Am J Neuroradiol. 2009 Mar;30(3):525-31
pubmed: 19147716
Magn Reson Med. 2017 Jan;77(1):398-410
pubmed: 26778056
IEEE Trans Med Imaging. 2003 Sep;22(9):1163-71
pubmed: 12956271
J Neurointerv Surg. 2020 Feb;12(2):156-164
pubmed: 31594798
JAMA. 2016 Sep 27;316(12):1279-88
pubmed: 27673305
Biomed Eng Online. 2015;14 Suppl 1:S11
pubmed: 25603312
Neuroradiology. 2018 Sep;60(9):889-901
pubmed: 30066278
Lancet Neurol. 2011 Oct;10(10):909-21
pubmed: 21939900
Stroke. 2009 Sep;40(9):3001-5
pubmed: 19590055
N Engl J Med. 2015 Mar 12;372(11):1009-18
pubmed: 25671797
Stroke. 2019 Oct;50(10):2790-2798
pubmed: 31495328
Stroke. 2018 Mar;49(3):e46-e110
pubmed: 29367334
J Neuroimaging. 2022 Sep;32(5):956-967
pubmed: 35838658
Int J Stroke. 2018 Oct;13(8):806-810
pubmed: 29956598
J Am Heart Assoc. 2015 Aug 06;4(8):e002289
pubmed: 26251284
Eur Radiol Exp. 2019 Feb 13;3(1):8
pubmed: 30758694
Neuroimage Clin. 2021;30:102573
pubmed: 33578323
Lancet. 2016 Apr 23;387(10029):1723-31
pubmed: 26898852
Stroke. 2021 May;52(5):1921-1928
pubmed: 33765866