Inconsistency of AI in intracranial aneurysm detection with varying dose and image reconstruction

Abstract Scanner-related changes in data quality are common in medical imaging, yet monitoring their impact on diagnostic AI performance remains challenging. In this study, we performed standardized consistency testing of an FDA-cleared and CE-marked AI for triage and notification of intracranial an...

Full description

Saved in:
Bibliographic Details
Main Authors: Leonie Goelz, Angelo Laudani, Ulrich Genske, Michael Scheel, Georg Bohner, Hans-Christian Bauknecht, Sven Mutze, Bernd Hamm, Paul Jahnke
Format: Article
Language:English
Published: Nature Portfolio 2025-06-01
Series:Scientific Reports
Subjects:
Online Access:https://doi.org/10.1038/s41598-025-04830-7
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849434313408380928
author Leonie Goelz
Angelo Laudani
Ulrich Genske
Michael Scheel
Georg Bohner
Hans-Christian Bauknecht
Sven Mutze
Bernd Hamm
Paul Jahnke
author_facet Leonie Goelz
Angelo Laudani
Ulrich Genske
Michael Scheel
Georg Bohner
Hans-Christian Bauknecht
Sven Mutze
Bernd Hamm
Paul Jahnke
author_sort Leonie Goelz
collection DOAJ
description Abstract Scanner-related changes in data quality are common in medical imaging, yet monitoring their impact on diagnostic AI performance remains challenging. In this study, we performed standardized consistency testing of an FDA-cleared and CE-marked AI for triage and notification of intracranial aneurysms across changes in image data quality caused by dose and image reconstruction. Our assessment was based on repeated examinations of a head CT phantom designed for AI evaluation, replicating a patient with three intracranial aneurysms in the anterior, middle and posterior circulation. We show that the AI maintains stable performance within the medium dose range but produces inconsistent results at reduced dose and, unexpectedly, at higher dose when filtered back projection is used. Data quality standards required for AI are stricter than those for neuroradiologists, who report higher aneurysm visibility rates and experience performance degradation only at substantially lower doses, with no decline at higher doses.
format Article
id doaj-art-03794751fcc64635a39da34c5f85cebb
institution Kabale University
issn 2045-2322
language English
publishDate 2025-06-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-03794751fcc64635a39da34c5f85cebb2025-08-20T03:26:42ZengNature PortfolioScientific Reports2045-23222025-06-011511910.1038/s41598-025-04830-7Inconsistency of AI in intracranial aneurysm detection with varying dose and image reconstructionLeonie Goelz0Angelo Laudani1Ulrich Genske2Michael Scheel3Georg Bohner4Hans-Christian Bauknecht5Sven Mutze6Bernd Hamm7Paul Jahnke8Department of Radiology and Neuroradiology, Unfallkrankenhaus BerlinDepartment of Radiology, Charité – Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, Berlin Institute of HealthDepartment of Radiology, Charité – Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, Berlin Institute of HealthDepartment of Neuroradiology, Charité – Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, Berlin Institute of HealthDepartment of Neuroradiology, Charité – Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, Berlin Institute of HealthDepartment of Neuroradiology, Charité – Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, Berlin Institute of HealthDepartment of Radiology and Neuroradiology, Unfallkrankenhaus BerlinDepartment of Radiology, Charité – Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, Berlin Institute of HealthDepartment of Radiology, Charité – Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, Berlin Institute of HealthAbstract Scanner-related changes in data quality are common in medical imaging, yet monitoring their impact on diagnostic AI performance remains challenging. In this study, we performed standardized consistency testing of an FDA-cleared and CE-marked AI for triage and notification of intracranial aneurysms across changes in image data quality caused by dose and image reconstruction. Our assessment was based on repeated examinations of a head CT phantom designed for AI evaluation, replicating a patient with three intracranial aneurysms in the anterior, middle and posterior circulation. We show that the AI maintains stable performance within the medium dose range but produces inconsistent results at reduced dose and, unexpectedly, at higher dose when filtered back projection is used. Data quality standards required for AI are stricter than those for neuroradiologists, who report higher aneurysm visibility rates and experience performance degradation only at substantially lower doses, with no decline at higher doses.https://doi.org/10.1038/s41598-025-04830-7Intracranial aneurysmArtificial intelligenceComputed tomography angiographyReproducibility of findingsImaging phantoms
spellingShingle Leonie Goelz
Angelo Laudani
Ulrich Genske
Michael Scheel
Georg Bohner
Hans-Christian Bauknecht
Sven Mutze
Bernd Hamm
Paul Jahnke
Inconsistency of AI in intracranial aneurysm detection with varying dose and image reconstruction
Scientific Reports
Intracranial aneurysm
Artificial intelligence
Computed tomography angiography
Reproducibility of findings
Imaging phantoms
title Inconsistency of AI in intracranial aneurysm detection with varying dose and image reconstruction
title_full Inconsistency of AI in intracranial aneurysm detection with varying dose and image reconstruction
title_fullStr Inconsistency of AI in intracranial aneurysm detection with varying dose and image reconstruction
title_full_unstemmed Inconsistency of AI in intracranial aneurysm detection with varying dose and image reconstruction
title_short Inconsistency of AI in intracranial aneurysm detection with varying dose and image reconstruction
title_sort inconsistency of ai in intracranial aneurysm detection with varying dose and image reconstruction
topic Intracranial aneurysm
Artificial intelligence
Computed tomography angiography
Reproducibility of findings
Imaging phantoms
url https://doi.org/10.1038/s41598-025-04830-7
work_keys_str_mv AT leoniegoelz inconsistencyofaiinintracranialaneurysmdetectionwithvaryingdoseandimagereconstruction
AT angelolaudani inconsistencyofaiinintracranialaneurysmdetectionwithvaryingdoseandimagereconstruction
AT ulrichgenske inconsistencyofaiinintracranialaneurysmdetectionwithvaryingdoseandimagereconstruction
AT michaelscheel inconsistencyofaiinintracranialaneurysmdetectionwithvaryingdoseandimagereconstruction
AT georgbohner inconsistencyofaiinintracranialaneurysmdetectionwithvaryingdoseandimagereconstruction
AT hanschristianbauknecht inconsistencyofaiinintracranialaneurysmdetectionwithvaryingdoseandimagereconstruction
AT svenmutze inconsistencyofaiinintracranialaneurysmdetectionwithvaryingdoseandimagereconstruction
AT berndhamm inconsistencyofaiinintracranialaneurysmdetectionwithvaryingdoseandimagereconstruction
AT pauljahnke inconsistencyofaiinintracranialaneurysmdetectionwithvaryingdoseandimagereconstruction