Deep learning based detection of intracranial aneurysms on digital subtraction angiography: A feasibility study.
Central nervous system
aneurysms
interventional
Journal
The neuroradiology journal
ISSN: 2385-1996
Titre abrégé: Neuroradiol J
Pays: United States
ID NLM: 101295103
Informations de publication
Date de publication:
Aug 2020
Aug 2020
Historique:
pubmed:
8
7
2020
medline:
1
5
2021
entrez:
8
7
2020
Statut:
ppublish
Résumé
Digital subtraction angiography is the gold standard for detecting and characterising aneurysms. Here, we assess the feasibility of commercial-grade deep learning software for the detection of intracranial aneurysms on whole-brain anteroposterior and lateral 2D digital subtraction angiography images. Seven hundred and six digital subtraction angiography images were included from a cohort of 240 patients (157 female, mean age 59 years, range 20-92; 83 male, mean age 55 years, range 19-83). Three hundred and thirty-five (47%) single frame anteroposterior and lateral images of a digital subtraction angiography series of 187 aneurysms (41 ruptured, 146 unruptured; average size 7±5.3 mm, range 1-5 mm; total 372 depicted aneurysms) and 371 (53%) aneurysm-negative study images were retrospectively analysed regarding the presence of intracranial aneurysms. The 2D data was split into testing and training sets in a ratio of 4:1 with 3D rotational digital subtraction angiography as gold standard. Supervised deep learning was performed using commercial-grade machine learning software (Cognex, ViDi Suite 2.0). Monte Carlo cross validation was performed. Intracranial aneurysms were detected with a sensitivity of 79%, a specificity of 79%, a precision of 0.75, a F1 score of 0.77, and a mean area-under-the-curve of 0.76 (range 0.68-0.86) after Monte Carlo cross-validation, run 45 times. The commercial-grade deep learning software allows for detection of intracranial aneurysms on whole-brain, 2D anteroposterior and lateral digital subtraction angiography images, with results being comparable to more specifically engineered deep learning techniques.
Sections du résumé
BACKGROUND
BACKGROUND
Digital subtraction angiography is the gold standard for detecting and characterising aneurysms. Here, we assess the feasibility of commercial-grade deep learning software for the detection of intracranial aneurysms on whole-brain anteroposterior and lateral 2D digital subtraction angiography images.
MATERIAL AND METHODS
METHODS
Seven hundred and six digital subtraction angiography images were included from a cohort of 240 patients (157 female, mean age 59 years, range 20-92; 83 male, mean age 55 years, range 19-83). Three hundred and thirty-five (47%) single frame anteroposterior and lateral images of a digital subtraction angiography series of 187 aneurysms (41 ruptured, 146 unruptured; average size 7±5.3 mm, range 1-5 mm; total 372 depicted aneurysms) and 371 (53%) aneurysm-negative study images were retrospectively analysed regarding the presence of intracranial aneurysms. The 2D data was split into testing and training sets in a ratio of 4:1 with 3D rotational digital subtraction angiography as gold standard. Supervised deep learning was performed using commercial-grade machine learning software (Cognex, ViDi Suite 2.0). Monte Carlo cross validation was performed.
RESULTS
RESULTS
Intracranial aneurysms were detected with a sensitivity of 79%, a specificity of 79%, a precision of 0.75, a F1 score of 0.77, and a mean area-under-the-curve of 0.76 (range 0.68-0.86) after Monte Carlo cross-validation, run 45 times.
CONCLUSION
CONCLUSIONS
The commercial-grade deep learning software allows for detection of intracranial aneurysms on whole-brain, 2D anteroposterior and lateral digital subtraction angiography images, with results being comparable to more specifically engineered deep learning techniques.
Identifiants
pubmed: 32633602
doi: 10.1177/1971400920937647
pmc: PMC7416354
doi:
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
311-317Références
Eur J Radiol. 2020 May;126:108925
pubmed: 32193036
Biomed Eng Online. 2019 Nov 14;18(1):110
pubmed: 31727057
Brain. 2000 Feb;123 ( Pt 2):205-21
pubmed: 10648430
J Neurosurg. 1997 Aug;87(2):176-83
pubmed: 9254079
Stroke. 1998 Jan;29(1):251-6
pubmed: 9445359
Neurosurg Focus. 2017 Jun;42(6):E12
pubmed: 28565977
AJNR Am J Neuroradiol. 2007 Jan;28(1):60-7
pubmed: 17213425
Radiology. 2004 Feb;230(2):510-8
pubmed: 14699177
Stroke. 2012 May;43(5):1309-14
pubmed: 22382160
Neuroradiology. 2004 Oct;46(10):842-50
pubmed: 15448952
N Engl J Med. 2006 Jan 26;354(4):387-96
pubmed: 16436770
J Vasc Interv Radiol. 2009 Jul;20(7 Suppl):S292-301
pubmed: 19560013
AJNR Am J Neuroradiol. 2018 Oct;39(10):1776-1784
pubmed: 29419402
J Clin Neurosci. 2007 Mar;14(3):252-5
pubmed: 17258133
J Neurosurg. 1999 Feb;90(2):207-14
pubmed: 9950490
AJNR Am J Neuroradiol. 2019 Jan;40(1):25-32
pubmed: 30573461
IEEE Trans Med Imaging. 2020 May;39(5):1448-1458
pubmed: 31689186
Lancet Neurol. 2011 Jul;10(7):626-36
pubmed: 21641282
J Neurointerv Surg. 2020 Apr;12(4):417-421
pubmed: 31444288
Radiology. 2020 Mar;294(3):487-489
pubmed: 31891322
AJNR Am J Neuroradiol. 2008 May;29(5):976-9
pubmed: 18258703
Radiology. 2014 May;271(2):553-60
pubmed: 24495263
AJNR Am J Neuroradiol. 2008 May;29(5):962-6
pubmed: 18258701
AJNR Am J Neuroradiol. 2002 May;23(5):768-71
pubmed: 12006274
Radiographics. 2017 Nov-Dec;37(7):2113-2131
pubmed: 29131760
Stroke. 2017 Aug;48(8):2105-2112
pubmed: 28667020
J Vasc Surg. 2007 Jan;45(1):149-54
pubmed: 17210400
Eur Radiol. 2005 Mar;15(3):441-7
pubmed: 15678323
Br J Radiol. 2018 Feb;91(1083):20170576
pubmed: 29215311
J Neurointerv Surg. 2020 May 29;:
pubmed: 32471827
Invest Radiol. 2017 Jul;52(7):434-440
pubmed: 28212138
J Magn Reson Imaging. 2018 Apr;47(4):948-953
pubmed: 28836310
Science. 2018 Dec 7;362(6419):1140-1144
pubmed: 30523106
Neurosurg Rev. 2011 Oct;34(4):409-16
pubmed: 21584689