Performance testing of a novel deep learning algorithm for the detection of intracranial hemorrhage and first trial under clinical conditions

Purpose: We evaluate the performance of a deep learning-based pipeline using a Dense U-net architecture for detection of intracranial hemorrhage (ICH) in unenhanced head computed tomography (CT) scans. Methods: A balanced database was assembled retrospectively, comprising a total of 872 CT scans (36...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Philipp Gruschwitz, Jan-Peter Grunz, Philipp Josef Kuhl, Aleksander Kosmala, Thorsten Alexander Bley, Bernhard Petritsch, Julius Frederik Heidenreich
Formato:	article
Lenguaje:	EN
Publicado:	Elsevier 2021
Materias:	Computed tomography Intracranial hemorrhage Artificial intelligence Neurosciences. Biological psychiatry. Neuropsychiatry RC321-571
Acceso en línea:	https://doaj.org/article/f1cda05677124302bb057aff68cffea7
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:f1cda05677124302bb057aff68cffea7
record_format	dspace
spelling	oai:doaj.org-article:f1cda05677124302bb057aff68cffea72021-12-03T04:01:38ZPerformance testing of a novel deep learning algorithm for the detection of intracranial hemorrhage and first trial under clinical conditions2772-528610.1016/j.neuri.2021.100005https://doaj.org/article/f1cda05677124302bb057aff68cffea72021-09-01T00:00:00Zhttp://www.sciencedirect.com/science/article/pii/S2772528621000054https://doaj.org/toc/2772-5286Purpose: We evaluate the performance of a deep learning-based pipeline using a Dense U-net architecture for detection of intracranial hemorrhage (ICH) in unenhanced head computed tomography (CT) scans. Methods: A balanced database was assembled retrospectively, comprising a total of 872 CT scans (362 with present ICH). Predictions by the algorithm were analyzed and compared to the radiology report (ground truth). Secondly, the algorithm's performance was tested in clinical environment: A total of 100 head CT scans (11 with present ICH) were analyzed simultaneously by the deep learning algorithm and a radiologist during clinical routine. The time until first temporary diagnosis of ICH was measured. Performances of the algorithm were evaluated in combination with the radiologist, when using it as triage tool. Results: In the retrospectively assembled dataset the deep learning algorithm detected ICH with a sensitivity of 91.4%, specificity of 90.4% and overall accuracy of 91.0%. In clinical environment, the algorithm was significantly faster compared to the temporary report of the assigned radiologist (24 ± 2 s vs. 613 ± 658 s, p < 0.001). When using the algorithm as a triage tool additional to the report of the assigned radiologist, a sensitivity of 100% was achieved. Conclusions: These results and the short processing time demonstrate the immense potential of deep learning applications for the use as triage tool and for additional review of manual reports.Philipp GruschwitzJan-Peter GrunzPhilipp Josef KuhlAleksander KosmalaThorsten Alexander BleyBernhard PetritschJulius Frederik HeidenreichElsevierarticleComputed tomographyIntracranial hemorrhageArtificial intelligenceNeurosciences. Biological psychiatry. NeuropsychiatryRC321-571ENNeuroscience Informatics, Vol 1, Iss 1, Pp 100005- (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Computed tomography Intracranial hemorrhage Artificial intelligence Neurosciences. Biological psychiatry. Neuropsychiatry RC321-571
spellingShingle	Computed tomography Intracranial hemorrhage Artificial intelligence Neurosciences. Biological psychiatry. Neuropsychiatry RC321-571 Philipp Gruschwitz Jan-Peter Grunz Philipp Josef Kuhl Aleksander Kosmala Thorsten Alexander Bley Bernhard Petritsch Julius Frederik Heidenreich Performance testing of a novel deep learning algorithm for the detection of intracranial hemorrhage and first trial under clinical conditions
description	Purpose: We evaluate the performance of a deep learning-based pipeline using a Dense U-net architecture for detection of intracranial hemorrhage (ICH) in unenhanced head computed tomography (CT) scans. Methods: A balanced database was assembled retrospectively, comprising a total of 872 CT scans (362 with present ICH). Predictions by the algorithm were analyzed and compared to the radiology report (ground truth). Secondly, the algorithm's performance was tested in clinical environment: A total of 100 head CT scans (11 with present ICH) were analyzed simultaneously by the deep learning algorithm and a radiologist during clinical routine. The time until first temporary diagnosis of ICH was measured. Performances of the algorithm were evaluated in combination with the radiologist, when using it as triage tool. Results: In the retrospectively assembled dataset the deep learning algorithm detected ICH with a sensitivity of 91.4%, specificity of 90.4% and overall accuracy of 91.0%. In clinical environment, the algorithm was significantly faster compared to the temporary report of the assigned radiologist (24 ± 2 s vs. 613 ± 658 s, p < 0.001). When using the algorithm as a triage tool additional to the report of the assigned radiologist, a sensitivity of 100% was achieved. Conclusions: These results and the short processing time demonstrate the immense potential of deep learning applications for the use as triage tool and for additional review of manual reports.
format	article
author	Philipp Gruschwitz Jan-Peter Grunz Philipp Josef Kuhl Aleksander Kosmala Thorsten Alexander Bley Bernhard Petritsch Julius Frederik Heidenreich
author_facet	Philipp Gruschwitz Jan-Peter Grunz Philipp Josef Kuhl Aleksander Kosmala Thorsten Alexander Bley Bernhard Petritsch Julius Frederik Heidenreich
author_sort	Philipp Gruschwitz
title	Performance testing of a novel deep learning algorithm for the detection of intracranial hemorrhage and first trial under clinical conditions
title_short	Performance testing of a novel deep learning algorithm for the detection of intracranial hemorrhage and first trial under clinical conditions
title_full	Performance testing of a novel deep learning algorithm for the detection of intracranial hemorrhage and first trial under clinical conditions
title_fullStr	Performance testing of a novel deep learning algorithm for the detection of intracranial hemorrhage and first trial under clinical conditions
title_full_unstemmed	Performance testing of a novel deep learning algorithm for the detection of intracranial hemorrhage and first trial under clinical conditions
title_sort	performance testing of a novel deep learning algorithm for the detection of intracranial hemorrhage and first trial under clinical conditions
publisher	Elsevier
publishDate	2021
url	https://doaj.org/article/f1cda05677124302bb057aff68cffea7
work_keys_str_mv	AT philippgruschwitz performancetestingofanoveldeeplearningalgorithmforthedetectionofintracranialhemorrhageandfirsttrialunderclinicalconditions AT janpetergrunz performancetestingofanoveldeeplearningalgorithmforthedetectionofintracranialhemorrhageandfirsttrialunderclinicalconditions AT philippjosefkuhl performancetestingofanoveldeeplearningalgorithmforthedetectionofintracranialhemorrhageandfirsttrialunderclinicalconditions AT aleksanderkosmala performancetestingofanoveldeeplearningalgorithmforthedetectionofintracranialhemorrhageandfirsttrialunderclinicalconditions AT thorstenalexanderbley performancetestingofanoveldeeplearningalgorithmforthedetectionofintracranialhemorrhageandfirsttrialunderclinicalconditions AT bernhardpetritsch performancetestingofanoveldeeplearningalgorithmforthedetectionofintracranialhemorrhageandfirsttrialunderclinicalconditions AT juliusfrederikheidenreich performancetestingofanoveldeeplearningalgorithmforthedetectionofintracranialhemorrhageandfirsttrialunderclinicalconditions
_version_	1718373964484444160

Performance testing of a novel deep learning algorithm for the detection of intracranial hemorrhage and first trial under clinical conditions

Ejemplares similares