Deep-learning based detection of vessel occlusions on CT-angiography in patients with suspected acute ischemic stroke

Abstract Swift diagnosis and treatment play a decisive role in the clinical outcome of patients with acute ischemic stroke (AIS), and computer-aided diagnosis (CAD) systems can accelerate the underlying diagnostic processes. Here, we developed an artificial neural network (ANN) which allows automate...

Full description

Saved in:
Bibliographic Details
Main Authors: Gianluca Brugnara, Michael Baumgartner, Edwin David Scholze, Katerina Deike-Hofmann, Klaus Kades, Jonas Scherer, Stefan Denner, Hagen Meredig, Aditya Rastogi, Mustafa Ahmed Mahmutoglu, Christian Ulfert, Ulf Neuberger, Silvia Schönenberger, Kai Schlamp, Zeynep Bendella, Thomas Pinetz, Carsten Schmeel, Wolfgang Wick, Peter A. Ringleb, Ralf Floca, Markus Möhlenbruch, Alexander Radbruch, Martin Bendszus, Klaus Maier-Hein, Philipp Vollmuth
Format: Article
Language:English
Published: Nature Portfolio 2023-08-01
Series:Nature Communications
Online Access:https://doi.org/10.1038/s41467-023-40564-8
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850179769784598528
author Gianluca Brugnara
Michael Baumgartner
Edwin David Scholze
Katerina Deike-Hofmann
Klaus Kades
Jonas Scherer
Stefan Denner
Hagen Meredig
Aditya Rastogi
Mustafa Ahmed Mahmutoglu
Christian Ulfert
Ulf Neuberger
Silvia Schönenberger
Kai Schlamp
Zeynep Bendella
Thomas Pinetz
Carsten Schmeel
Wolfgang Wick
Peter A. Ringleb
Ralf Floca
Markus Möhlenbruch
Alexander Radbruch
Martin Bendszus
Klaus Maier-Hein
Philipp Vollmuth
author_facet Gianluca Brugnara
Michael Baumgartner
Edwin David Scholze
Katerina Deike-Hofmann
Klaus Kades
Jonas Scherer
Stefan Denner
Hagen Meredig
Aditya Rastogi
Mustafa Ahmed Mahmutoglu
Christian Ulfert
Ulf Neuberger
Silvia Schönenberger
Kai Schlamp
Zeynep Bendella
Thomas Pinetz
Carsten Schmeel
Wolfgang Wick
Peter A. Ringleb
Ralf Floca
Markus Möhlenbruch
Alexander Radbruch
Martin Bendszus
Klaus Maier-Hein
Philipp Vollmuth
author_sort Gianluca Brugnara
collection DOAJ
description Abstract Swift diagnosis and treatment play a decisive role in the clinical outcome of patients with acute ischemic stroke (AIS), and computer-aided diagnosis (CAD) systems can accelerate the underlying diagnostic processes. Here, we developed an artificial neural network (ANN) which allows automated detection of abnormal vessel findings without any a-priori restrictions and in <2 minutes. Pseudo-prospective external validation was performed in consecutive patients with suspected AIS from 4 different hospitals during a 6-month timeframe and demonstrated high sensitivity (≥87%) and negative predictive value (≥93%). Benchmarking against two CE- and FDA-approved software solutions showed significantly higher performance for our ANN with improvements of 25–45% for sensitivity and 4–11% for NPV (p ≤ 0.003 each). We provide an imaging platform ( https://stroke.ccibonn.ai/ ) for online processing of medical imaging data with the developed ANN, including provisions for data crowdsourcing, which will allow continuous refinements and serve as a blueprint to build robust and generalizable AI algorithms.
format Article
id doaj-art-24a3d137218a42cfa5b7e7a9db4adb74
institution OA Journals
issn 2041-1723
language English
publishDate 2023-08-01
publisher Nature Portfolio
record_format Article
series Nature Communications
spelling doaj-art-24a3d137218a42cfa5b7e7a9db4adb742025-08-20T02:18:25ZengNature PortfolioNature Communications2041-17232023-08-0114111510.1038/s41467-023-40564-8Deep-learning based detection of vessel occlusions on CT-angiography in patients with suspected acute ischemic strokeGianluca Brugnara0Michael Baumgartner1Edwin David Scholze2Katerina Deike-Hofmann3Klaus Kades4Jonas Scherer5Stefan Denner6Hagen Meredig7Aditya Rastogi8Mustafa Ahmed Mahmutoglu9Christian Ulfert10Ulf Neuberger11Silvia Schönenberger12Kai Schlamp13Zeynep Bendella14Thomas Pinetz15Carsten Schmeel16Wolfgang Wick17Peter A. Ringleb18Ralf Floca19Markus Möhlenbruch20Alexander Radbruch21Martin Bendszus22Klaus Maier-Hein23Philipp Vollmuth24Department of Neuroradiology, Heidelberg University HospitalDivision of Medical Image Computing, German Cancer Research Center (DKFZ)Department of Neuroradiology, Heidelberg University HospitalDepartment of Neuroradiology, Bonn University HospitalDivision of Medical Image Computing, German Cancer Research Center (DKFZ)Division of Medical Image Computing, German Cancer Research Center (DKFZ)Division of Medical Image Computing, German Cancer Research Center (DKFZ)Department of Neuroradiology, Heidelberg University HospitalDepartment of Neuroradiology, Heidelberg University HospitalDepartment of Neuroradiology, Heidelberg University HospitalDepartment of Neuroradiology, Heidelberg University HospitalDepartment of Neuroradiology, Heidelberg University HospitalNeurology Clinic, Heidelberg University HospitalDepartment of Diagnostic and Interventional Radiology with Nuclear Medicine, Thoraxklinik at University of HeidelbergDepartment of Neuroradiology, Bonn University HospitalInstitute for Applied Mathematics, University of BonnDepartment of Neuroradiology, Bonn University HospitalNeurology Clinic, Heidelberg University HospitalNeurology Clinic, Heidelberg University HospitalDivision of Medical Image Computing, German Cancer Research Center (DKFZ)Department of Neuroradiology, Heidelberg University HospitalDepartment of Neuroradiology, Bonn University HospitalDepartment of Neuroradiology, Heidelberg University HospitalDivision of Medical Image Computing, German Cancer Research Center (DKFZ)Department of Neuroradiology, Heidelberg University HospitalAbstract Swift diagnosis and treatment play a decisive role in the clinical outcome of patients with acute ischemic stroke (AIS), and computer-aided diagnosis (CAD) systems can accelerate the underlying diagnostic processes. Here, we developed an artificial neural network (ANN) which allows automated detection of abnormal vessel findings without any a-priori restrictions and in <2 minutes. Pseudo-prospective external validation was performed in consecutive patients with suspected AIS from 4 different hospitals during a 6-month timeframe and demonstrated high sensitivity (≥87%) and negative predictive value (≥93%). Benchmarking against two CE- and FDA-approved software solutions showed significantly higher performance for our ANN with improvements of 25–45% for sensitivity and 4–11% for NPV (p ≤ 0.003 each). We provide an imaging platform ( https://stroke.ccibonn.ai/ ) for online processing of medical imaging data with the developed ANN, including provisions for data crowdsourcing, which will allow continuous refinements and serve as a blueprint to build robust and generalizable AI algorithms.https://doi.org/10.1038/s41467-023-40564-8
spellingShingle Gianluca Brugnara
Michael Baumgartner
Edwin David Scholze
Katerina Deike-Hofmann
Klaus Kades
Jonas Scherer
Stefan Denner
Hagen Meredig
Aditya Rastogi
Mustafa Ahmed Mahmutoglu
Christian Ulfert
Ulf Neuberger
Silvia Schönenberger
Kai Schlamp
Zeynep Bendella
Thomas Pinetz
Carsten Schmeel
Wolfgang Wick
Peter A. Ringleb
Ralf Floca
Markus Möhlenbruch
Alexander Radbruch
Martin Bendszus
Klaus Maier-Hein
Philipp Vollmuth
Deep-learning based detection of vessel occlusions on CT-angiography in patients with suspected acute ischemic stroke
Nature Communications
title Deep-learning based detection of vessel occlusions on CT-angiography in patients with suspected acute ischemic stroke
title_full Deep-learning based detection of vessel occlusions on CT-angiography in patients with suspected acute ischemic stroke
title_fullStr Deep-learning based detection of vessel occlusions on CT-angiography in patients with suspected acute ischemic stroke
title_full_unstemmed Deep-learning based detection of vessel occlusions on CT-angiography in patients with suspected acute ischemic stroke
title_short Deep-learning based detection of vessel occlusions on CT-angiography in patients with suspected acute ischemic stroke
title_sort deep learning based detection of vessel occlusions on ct angiography in patients with suspected acute ischemic stroke
url https://doi.org/10.1038/s41467-023-40564-8
work_keys_str_mv AT gianlucabrugnara deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT michaelbaumgartner deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT edwindavidscholze deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT katerinadeikehofmann deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT klauskades deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT jonasscherer deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT stefandenner deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT hagenmeredig deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT adityarastogi deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT mustafaahmedmahmutoglu deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT christianulfert deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT ulfneuberger deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT silviaschonenberger deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT kaischlamp deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT zeynepbendella deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT thomaspinetz deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT carstenschmeel deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT wolfgangwick deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT peteraringleb deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT ralffloca deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT markusmohlenbruch deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT alexanderradbruch deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT martinbendszus deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT klausmaierhein deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke
AT philippvollmuth deeplearningbaseddetectionofvesselocclusionsonctangiographyinpatientswithsuspectedacuteischemicstroke