An anomaly detection approach to identify chronic brain infarcts on MRI

Abstract The performance of current machine learning methods to detect heterogeneous pathology is limited by the quantity and quality of pathology in medical images. A possible solution is anomaly detection; an approach that can detect all abnormalities by learning how ‘normal’ tissue looks like. In...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Kees M. van Hespen, Jaco J. M. Zwanenburg, Jan W. Dankbaar, Mirjam I. Geerlings, Jeroen Hendrikse, Hugo J. Kuijf
Formato: article
Lenguaje:EN
Publicado: Nature Portfolio 2021
Materias:
R
Q
Acceso en línea:https://doaj.org/article/5bc99581278d428aa446e930a860e604
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:5bc99581278d428aa446e930a860e604
record_format dspace
spelling oai:doaj.org-article:5bc99581278d428aa446e930a860e6042021-12-02T14:37:39ZAn anomaly detection approach to identify chronic brain infarcts on MRI10.1038/s41598-021-87013-42045-2322https://doaj.org/article/5bc99581278d428aa446e930a860e6042021-04-01T00:00:00Zhttps://doi.org/10.1038/s41598-021-87013-4https://doaj.org/toc/2045-2322Abstract The performance of current machine learning methods to detect heterogeneous pathology is limited by the quantity and quality of pathology in medical images. A possible solution is anomaly detection; an approach that can detect all abnormalities by learning how ‘normal’ tissue looks like. In this work, we propose an anomaly detection method using a neural network architecture for the detection of chronic brain infarcts on brain MR images. The neural network was trained to learn the visual appearance of normal appearing brains of 697 patients. We evaluated its performance on the detection of chronic brain infarcts in 225 patients, which were previously labeled. Our proposed method detected 374 chronic brain infarcts (68% of the total amount of brain infarcts) which represented 97.5% of the total infarct volume. Additionally, 26 new brain infarcts were identified that were originally missed by the radiologist during radiological reading. Our proposed method also detected white matter hyperintensities, anomalous calcifications, and imaging artefacts. This work shows that anomaly detection is a powerful approach for the detection of multiple brain abnormalities, and can potentially be used to improve the radiological workflow efficiency by guiding radiologists to brain anomalies which otherwise remain unnoticed.Kees M. van HespenJaco J. M. ZwanenburgJan W. DankbaarMirjam I. GeerlingsJeroen HendrikseHugo J. KuijfNature PortfolioarticleMedicineRScienceQENScientific Reports, Vol 11, Iss 1, Pp 1-10 (2021)
institution DOAJ
collection DOAJ
language EN
topic Medicine
R
Science
Q
spellingShingle Medicine
R
Science
Q
Kees M. van Hespen
Jaco J. M. Zwanenburg
Jan W. Dankbaar
Mirjam I. Geerlings
Jeroen Hendrikse
Hugo J. Kuijf
An anomaly detection approach to identify chronic brain infarcts on MRI
description Abstract The performance of current machine learning methods to detect heterogeneous pathology is limited by the quantity and quality of pathology in medical images. A possible solution is anomaly detection; an approach that can detect all abnormalities by learning how ‘normal’ tissue looks like. In this work, we propose an anomaly detection method using a neural network architecture for the detection of chronic brain infarcts on brain MR images. The neural network was trained to learn the visual appearance of normal appearing brains of 697 patients. We evaluated its performance on the detection of chronic brain infarcts in 225 patients, which were previously labeled. Our proposed method detected 374 chronic brain infarcts (68% of the total amount of brain infarcts) which represented 97.5% of the total infarct volume. Additionally, 26 new brain infarcts were identified that were originally missed by the radiologist during radiological reading. Our proposed method also detected white matter hyperintensities, anomalous calcifications, and imaging artefacts. This work shows that anomaly detection is a powerful approach for the detection of multiple brain abnormalities, and can potentially be used to improve the radiological workflow efficiency by guiding radiologists to brain anomalies which otherwise remain unnoticed.
format article
author Kees M. van Hespen
Jaco J. M. Zwanenburg
Jan W. Dankbaar
Mirjam I. Geerlings
Jeroen Hendrikse
Hugo J. Kuijf
author_facet Kees M. van Hespen
Jaco J. M. Zwanenburg
Jan W. Dankbaar
Mirjam I. Geerlings
Jeroen Hendrikse
Hugo J. Kuijf
author_sort Kees M. van Hespen
title An anomaly detection approach to identify chronic brain infarcts on MRI
title_short An anomaly detection approach to identify chronic brain infarcts on MRI
title_full An anomaly detection approach to identify chronic brain infarcts on MRI
title_fullStr An anomaly detection approach to identify chronic brain infarcts on MRI
title_full_unstemmed An anomaly detection approach to identify chronic brain infarcts on MRI
title_sort anomaly detection approach to identify chronic brain infarcts on mri
publisher Nature Portfolio
publishDate 2021
url https://doaj.org/article/5bc99581278d428aa446e930a860e604
work_keys_str_mv AT keesmvanhespen ananomalydetectionapproachtoidentifychronicbraininfarctsonmri
AT jacojmzwanenburg ananomalydetectionapproachtoidentifychronicbraininfarctsonmri
AT janwdankbaar ananomalydetectionapproachtoidentifychronicbraininfarctsonmri
AT mirjamigeerlings ananomalydetectionapproachtoidentifychronicbraininfarctsonmri
AT jeroenhendrikse ananomalydetectionapproachtoidentifychronicbraininfarctsonmri
AT hugojkuijf ananomalydetectionapproachtoidentifychronicbraininfarctsonmri
AT keesmvanhespen anomalydetectionapproachtoidentifychronicbraininfarctsonmri
AT jacojmzwanenburg anomalydetectionapproachtoidentifychronicbraininfarctsonmri
AT janwdankbaar anomalydetectionapproachtoidentifychronicbraininfarctsonmri
AT mirjamigeerlings anomalydetectionapproachtoidentifychronicbraininfarctsonmri
AT jeroenhendrikse anomalydetectionapproachtoidentifychronicbraininfarctsonmri
AT hugojkuijf anomalydetectionapproachtoidentifychronicbraininfarctsonmri
_version_ 1718391007218761728